 Should start recording and I'm gonna go to OBS and see that that's actually true It did start recording fantastic, and I'll cut this out before I upload the video All right. Hello everyone. Welcome back to another stream It's been a long time since last time because I have been moving as you can see the room is completely different now My chair is different. The picture is the same. I brought that So this stream is not gonna be a rust specific stream It's not going to be a coding stream instead. It's just going to be a Q&A stream. There are enough things that have changed in In my life since last time that I figured a Q&A is a good thing to do just because people probably have questions Things have changed in the world too If you have submitted questions in advance, then you can watch them being answered now You can also ask questions if you're watching this live You can ask questions during the stream the link to where you can ask questions and vote on questions is in the video Description or someone also just posted it in chat. They're both the same. They both work What I'll do is I'll go through the list of questions that have come in You know by how much they've been voted up And I'll skip anyone that I don't care about I don't think I'll skip very many And I'll just go down the list and what we'll see how far we get there are a lot of questions So, you know, this is a good place to start Okay question number one first question in Norway first question on a stream in Norway How did you find your new position or did they find you? So for a little bit of context, I recently left AWS as part of my move and I joined a company called Helsing and That move there's all sorts of things Buried in that move But in terms of the specific questions of how I find that my new position what happened was when I Decided that I was probably going to be leaving AWS. I posted to Twitter to mast it on and to LinkedIn saying I'm Likely going to be leaving my position at AWS. I am moving to Norway. I'm looking for work here are the sort of Not criteria, but rather here's what I know. Here's what I'm good at. Here's what I'm looking for and I'm in a slightly privileged position here, right because Normally when you apply to jobs, you can't really just state your requirements in my case I had the luxury of being able to do that, which was really nice And so I said but things like, you know, I don't want to work on anything. That's Bitcoin related for example, which is just that's my preference here And you know, these are industries I might be interested in and I got a bunch of emails in response and Twitter DMs and whatnot Helsing was not one of those initially Or rather what happened was I think a few weeks later I got an email from someone who had been watching my streams for a while being like, oh, I noticed you're looking for work I work at this company. Would you be interested in talking to them? And I was like, sure that sounds interesting. Why not? You know, it looks like a company that does does does interesting things in a domain that I'm not used to and so I set up a Call with one of the hiring managers there and then it sort of ran from there So in a sense it was kind of both, but mostly they found me through this person who'd been watching my videos Okay, next question, let's see if there are any follow-ups to that Nope, no follow-ups. So this is a sensitive question I believe salaries in the EU are not aligned with North America and especially the US Is there a big gap between your new and old one? It's it's not a sensitive question or rather, I think decent chunks of the world think that it's a sensitive question I don't think it should be I think salary is one of those things that it's extremely important to be pretty transparent about Because otherwise you end up with immense pay gaps both between regions between people between Companies and I think I think it benefits all of us to be relatively transparent with with our salaries some of that is complicated Depending on the company then depending on the country you work in and the contract you've signed But what I'll say is that So at Amazon and I've said this before my salary sort of total compensation including stocks and the like Was around $450,000 a year Towards the end But in reality it was actually a fair bit lower because of the drop in Amazon stock price So it was almost closer to 350 by the end Which is pretty crazy that that's abnormal right normally they pay a decent amount more And I was also on the sort of promotion tractor principal engineer, which would have bumped that pay By some amount as well, but so I was exiting let's say around four hundred thousand dollars a year the I didn't end up with an actual offer from Amazon in Europe But my understanding is that the offer would be Significantly lower in Europe than it would be in the US that also varies by company. So at Meta for example This is only based on my understanding The the pay in Europe is lower than the US but not by that much It's like 10 to 20 percent and Amazon. I think the pay cut is about 50 percent. Maybe even more It's just pretty significant Helsing pays pretty competitively the challenge that any company that's not On the stock market yet has is that when they pay you in stocks Those stocks aren't worth anything or rather that they have an estimated worth like they can estimate the value of the company based on you know valuations and and and seed rounds and the like but but you they're not liquid like you can't sell them and get Money so it is a little bit of a bet and so it depends on whether you consider those if you're given stocks Whether you consider them being worth anything So the the cash compensation that I'm getting from Helsing is Competitive with what I got at Amazon. So it's about 250,000 euros a year and then there's stocks on top of that which are worth Question mark amount of dollars. I think the sort of target was that the total comp would be about the same given the current valuation of the company What's sort of the expectation that the valuation of the company might go up over time? But that's roughly where we're at So in that sense, there's not actually a big pay gap between my old position and my new position Which is pretty impressive given that I'm in Europe This is one of the things that I've gotten the impression that Helsing is doing pretty intentionally is that they Understand that this is a place where it's important to attract Very good engineers and that means that you have to pay competitive salaries And That's certainly what I got You know, it depends also what you compare with right if you compare with like a I don't know high-frequency trading firm in the US then no, they're not competitive, but but You're comparing very different industries as well Follow-up questions Not any follow-up questions yet. All right I'll try to take follow-up questions before I move on to the next one Just because although looping back is annoying, especially for the people watching like video on demand afterwards because then I put I Put like chapter headlines for each question and then it becomes a new chapter headline that refers to your previous question It just gets annoying Did you consider the stock value at what price when evaluating the offer? Yes, I did so the offer that I got from Helsing included some number of stocks and The sort of estimation of the current value of the company and I certainly took that into account The tricky part right is that it's not about what the company is currently worth Which you can estimate pretty well based on seed seed rounds You have to guess what they're going to be worth by the time the stocks are liquid And that's much harder You you can sort of guess based on the growth of the company and what kind of contracts they're landing and stuff But it's a very it's a gamble But I did take that into account. Yes When you ask questions, please ask them in that special Q&A link that's in chat and in the video description Next question, what's your take on the whole rust foundation conundrum? Okay, I Knew this question was gonna come arguably I should have prepared better for it, but here's the deal This is the the debate around the rust foundation and the the rust trademark policy So my take here is that the rust community Wants to have their cake and eat it too The rust community both wants the foundation to be fully transparent about everything it does and and sort of Show all of their work the entirety of the way and include the community every step of the process But at the same time they expect everything that comes out of the foundation to be perfect the first time around Or rather if it's not perfect if it's if there's anything at all wrong then thus the amount of Not scrutiny because scrutiny is not problematic, but the amount of criticism is get is immense and that's Unfortunate Because you end up in a position where the foundation can't win there's not the foundation can't like either way they go They're gonna get severely criticized And that is gonna demoralize any effort that goes into it and everyone who's involved in that process And I think the community is doing itself a disservice here. I think it is very important as a community that we are Diligent about making sure that the what the foundation does is aligned with what we believe that rust as a community needs and should have And the direction that we should go but at the same time we should also recognize that it takes work and like actual people being paid work hours And and insight and iteration to get to something that's really good and that means you're going to see things that aren't perfect from the beginning That all said I do think that the Way that this rolled out from the foundation of the communication around it could have been better to avoid the kind of I don't know and misreading is maybe wrong but That The misunderstanding that happened if you will I think the communication there could have been better but Ultimately, I think this is actually a problem with how the community looks at the foundation and thinks of the foundation I See someone is gaming the system. So I'm gonna hide their question Okay, follow-up questions on the rust foundation For I'll also add that I do think that that trademark policy was problematic in a number of ways as has been described that I'm not not questioning that at all it's more that This was clearly a draft that was expected to be iterated on and I think we should give the the foundation some leeway to Propose things that aren't perfect Okay, next question What could possibly be a reasonable application of AI in the military? Helsing's website is not remotely clear about what they actually do an AI for military applications seems like a gigantic Mindfield of serious concerns a computer cannot be held accountable. Therefore a computer must never make management slash military decisions I'm I don't love the phrasing of this question But but I think the core of it is a good one Which is like why would you want AI in the military or near anything that is as Serious and potentially deadly as what military Military involvement entails and the the answer here is that and this is personal opinion just to be clear My take here is that it's actually critical to get I AI into these situations because there is such an Overload of information. It's a fire hose of information once you start connecting all these different systems And so it's not about using AI to make decisions I agree with you there that if you take the human out of the loop here, then you run into really problematic territory But but that's also not Helsing's goal as far as I've deduced or that I understand from talking to the people there Instead the goal is to use AI to make the humans decision-making process and the humans Insight into what happens in a war zone or on a battlefield To give them better insight better understanding better overview so that they can make better decisions And there I think it is hugely valuable because there's a there's a limit to how we as Humans can like scale up our information processing capabilities. We need a Hierarchy of information, right? We need things to be summarized at different levels and be brought up to us in a way in a in a Format that we can make decisions on if you are watching over a battlefield and You just get all of the information about all of the assets that are there friend or foe Then you're gonna be overwhelmed You're not gonna be able to take intelligent decisions because you're not gonna see the patterns and the hope is that by using AI here You can try to distill the things that are important surface those better And then the human can make decisions based on that so it's not about making AI make decisions It is about making AI let humans make better decisions Okay follow up questions There is no best choice There is no best choice um, I Don't know what that's in reference to but I actually think that there it's true that there's no right choice very often in Any conflict really? But but certainly in war But but I do think that there are better and worse decisions and the goal here is to allow the humans to make better decisions And I don't think we want AI to be making decisions that there might be some exceptions right there might be cases where The it's not tenable for a human to make decision because Like I don't know you get a I can't think of an example of the top of my head but but I know that there are some but The goal shouldn't be to have the AI make decisions and I think any time where there's a Potential for the AI to make make a decision. You have to be extremely rigorous about Why are we allowing this to be the case? What are the options? Is this actually something we want to do? I'm worried it's one step away from making from making AI make decisions. I mean, that's true kind of right so There's a there's a big difference to whether you're using AI for Information gathering and summarization then if you're using AI to send decisions down though those are two very different pathways and I don't think that having AI in one makes it easier to do AI in the other Maybe there's a question of like having technology built in here at all, but but I think they're actually pretty disconnected So I'm not too worried about like a slippery slope here Lazy humans though is a problem and this is something that I know Helsing is thinking pretty actively about as well that you want to make sure that the The systems that you build that surface information Don't try to like Make a suggested decision that the human then clicks okay on you want to make sure that it's it's it's actually a human in the loop The human in a loop is not just a yes button clicker And and that that goes through the entire process right that that depends on how you design your ui's It's an entire UX issue. It's like it's almost Psychology if you will and so so point very much taken the lazy humans is something you need to worry about But but at the same time I think that is a problem. You have to worry about regardless of how you set up these systems Okay, I'm gonna move on to the next question Does the tech stack used for AI? Oh, is the tech stack used for AI in rust as well We'd love to know more on where rust helps in the AI space and the products built So one fun part here is I don't yet have access to the Helsing code base because I don't actually start until August So I can't tell you anything about the actual code But I can tell you about sort of roughly what I know about the the structure of the the or the Architecture of you will in the design of the software stack a lot of the AI stuff is in Python The hope is that we can use rust to make those parts better And maybe over time we end up like moving a lot of that to rust or at least what makes sense at the same time I don't think the goal should be rust everywhere. I think for example when you have AI researchers working interactively on a problem I don't think you want that to be in rust I think you you want the the higher turnaround the the shorter, you know build debug run cycles For that which Python gives you and then maybe you want to translate the things that are actually going to go into production into rust So I don't think the goal is to remove Python altogether I think that is a I think that's the wrong goal to have in mind instead the question is where can we use rust to make this This overall system better now as for where well if Helsing does AI then where is the rust and the answer is the AI is is really mostly present in the sort of information Collation space that's where the AI shines But there are all these underlying systems that are used to disseminate information gather information Coordinate between different units make essentially building the communication system And that's basically all in rust so I think you know the network stack You need to talk through some really weird like hardware and military equipment like a lot of this stuff is very low-level Things you have really intermittent network connectivity because that plane isn't near that radar anymore, right? So links come and go. They're really unreliable. They're really Low bandwidth. They're very high latency. Maybe all of the above And so you need to build a solid tech stack for even just exchanging information in that kind of severely distributed system and Those bits are all in rust same things with everything that then gathers the information and brings it back to where it can be analyzed Anything that sends, you know control operations back Essentially everything except for the AI which turns out to be a fairly substantial bit of distributed systems engineering so so There isn't too much rust in the AI part of the stack, but I'm hoping we can get there more and more specifically because You know, you really want low latency for anything that's being displayed to humans, right? And their Python doesn't really shine anymore. You end up with pretty high turnaround times So once you have encoded the algorithm Having an easy way to then turn that into the sort of lower latency higher throughput rust front-end bits as well Would be beneficial Does rust make sense here ML nowadays is mostly glue and I don't think rust can be Python at that Yes, so that's sort of the discussion, right? Which is I think you're right that for the glue code Python is pretty great for anything. That's changing rapidly like for anything That's you know, AI researchers anything that is Essentially prototyping but once you've sort of solidified that this is how this thing is going to work now I think it makes a lot more sense to have it be in rust because even though it's glue It ends up being on the critical path for a lot of things and so making it be in rust is probably a worthwhile investment Oh, yeah, securing information at at rest All of the storage systems for things like the AI models Getting that data exchanging it to where the models need to be run between where it needs to be trained and where it needs to be stored All of that is also in rust Okay Next question What is your skincare routine women youtubers are always asked this so men should get to answer to it's a great question I was also told by my my partner that I have to answer it, which I was planning to anyway My skincare routine. I think as my girlfriend would say is Suboptimal, you know, there's a bunch more things that I probably should be doing I Get pretty dry hands, so I need to use like moisturizer a bunch and I don't use it as much as I should And then I get, you know, very coarse hands. It's not not great I Don't do anything super special Otherwise what I do do is as you as you may have seen sometimes Sometimes I fully shave and everyone thinks it's really weird but it's just nice sometimes just like let my skin breathe and then I put on like Moisturizers like phase See, I didn't even know that there were special moisturizers for faces But there are and so when I shave I put those on and everything gets gets gets nice And then because it gets it can get pretty the skin under the beard can get pretty dry over time I Think that's really about it which is I Wish I had more to share on this point because I feel like it would be interesting But I don't really which just means that I'm underperforming in this area Oh Beard oil I used beard oil for a little bit But it didn't really work well for me because I think my skin when my skin needs isn't oil It's whatever the opposite of oil is like There's a there's like two things and you don't want to use one if you have skin that needs the other But yeah, I tried using it for a while and that just didn't really do it for me and just left my skin kind of greasy So I wasn't that wasn't a huge fan But I tried it didn't make it meaningfully softer either Okay, next question Did your not a riot II make and make interviewing a breeze the size of the company have an impact on that I'm thinking big companies are more rigid So it really varies by company and You know the fact that I have a lot of Not just that I'm known that that That carries some weight, but also the fact that I've done a lot of live coding So people have seen me program right right like it becomes a little weird to do something like a Coding question in an interview when they've seen me program live right because it feels like that's already kind of covered but that said some companies are a lot more rigid than others and The it's actually not clear that it directly Correlates with the size of the company. I had some decently large companies for whom the interview process for me was different That didn't mean I got to skip a lot of steps It just meant that I went in a slightly different bucket where I talked to different people and about different kinds of things And maybe that's more that ended up being maybe a more senior hiring path or something That's hard for me to gauge from the outside the in terms of the In terms of the smaller companies some of them were just like we want to hire you that there was no Interviewing really except they reached out and we're like we want to hire you and that's nice But at the same time I do think it's Worth while to interview people because even if you just have like one conversation with them I guess if your company is like two people then then sure it doesn't matter But if you're like 20 people it's worth talking to them a little bit more to get a A bit more of an insight into what they're like as a person because you know what when you see me on stream I think that's pretty close to how I behave normally But that's different to if you're having a direct conversation with me Like the streams aren't back and forth. They're they're sort of thought streams or dumps for my brain And that's not normally how you would interact with a co-worker So I do think it's worthwhile having some of those interview conversations. Maybe you skip the technical interview, but like So so it in reality the answer here is that being known meant that Most of the companies I talked to came in with the either the expectation that the coding part was a formality or just with a We're not even gonna do coding And then for some of them it went all the way to we just want to hire you and in some sense That's that's kind of a red flag for me Like I want to make sure that they have a conversation with me because otherwise it means that they're not Sort of to use some Amazon phrasing here diving deep on the candidates They're getting in to figure out whether they are actually someone they want to work with and not just someone They think is a good engineer in terms of The companies that are sort of mid-sized there it varied a lot For some of them I went through a completely normal interview loop, but all of it with a weird air of like Yes, we apologize for the inconvenience, which like I don't mind I understand why you have to follow a process and it's fairer that way. So don't apologize To some companies that were just like, okay We're not going to do the standard process because it doesn't really make sense for you but we do want to have these conversations and and to figure out whether you fit with the company and Ultimately like for Helsing, for example, that's what I went through So I didn't go through like a normal coding interview for example But I did talk to a bunch of senior people and have those conversations of like What do I think about software engineering? How do we interact when we have a discussion? And and so I think that ended up being a nice middle ground So it I don't know that it makes interviewing a breeze, but it makes it different It also depends what you're looking for like if I was just looking for a job Then yes, it makes it a breeze because I could just pick one of the one companies that just wanted to hire me but in terms of The other companies for the ones that have the standard process those are also, you know, there's no difference There's no breeze or no breeze but for the ones where you go through like a special path. It's I don't want to say intimidating, but it is It's harder to prepare for because at that point you're just having conversations with people which Isn't necessarily Good or bad, but it's just it's not a it doesn't feel like a formal interview setting and so it's always a little hard to figure out how you should present Like what voice you should use if that makes sense So it just it made interviewing weird and different Was there correlation between company size and compensation a little bit the smaller companies can't offer quite as much usually, but the biggest difference is actually whether they are a publicly traded company or not so if If the stock grants they give you have actual liquid value or not Because what a lot of those startups tend to do is that because they can't pay you as much They pay you the the sort of gap in stocks based on their guesstimated value of those stocks But that's not super helpful. It's not that's not real compensation, right? It's not money I can take to buy food with And and so in that sense, you know that the the pays are correlated in the sense that The bigger companies are more likely to be public and so therefore you have more liquid money coming out of it But at the same time if you if you take the startups at face value in terms of their valuations and what the stocks are actually worth Then you get to a point where actually They're decently competitive It also depends on the level so so the other thing I found was that the the smaller companies tend to cap out sooner Like it's rare that they're looking for like a principal engineer because they don't really need one if they have five people Or they might want one but it's gonna be harder for them to attract one So if you're coming as like a principal engineer or a senior principal engineer from one of the big tech companies into a startup your pay is going to drop because the those companies just have a sort of Ceiling around like the senior engineer senior staff like somewhere around there That's sort of where they top off unless you then become like the head of engineering or VP of engineering or CTO or something And There are some very well-funded startups That's also true and some of them the other thing that's interesting is some companies pay because of the industry They're on or the reputation they have so if they have a worse reputation or they're in a more Controversial industry then they pay more because they have to in order to attract candidates What's your current title at Helsing my title currently is a software engineer, we're still basically working out what the Ladder looks like so to speak so I'm guessing that's gonna change and probably not too long For anyone who's asking questions, please ask them in the question page So there's one in the video description and someone can please post it in the chat as well Where you can go and ask questions vote on questions use that instead because the chat disappears so quickly Thank you. Wow. That's a lot of copies of the same thing Yeah, and then there are a lot of like Bitcoin web three companies to try to pay you in crypto And I'm just not interested in that. I don't believe it has value Okay, next question does the lack of Sun in northern Europe affect your mood compared to LA Yes, but positively I like darkness I I mean currently because we're in the middle of summer. There's actually more sun here there and then there is an LA But in winter, of course, there's less sun, but like now The the sun only turned I want to say a month ago or something. So we've had like the sun sets that Well, like 1 2 a.m. And then gets up at like Five or six like it's a lot of sun But then in winter, of course, it's you have in Oslo, what five hours of sun for four or five No, I think about five hours of sunlight So, you know, the difference is pretty stark and I I like the variation and generally I like darkness So I like the winter a lot and I don't really miss LA In any way really but but especially when it comes to the Sun my big problem with LA was not the amount of hours of Sun but the fact that the Sun is so strong like going outside when it's really sunny You're just like you burn you feel very hot. It's just not very nice So the generally the lack of Sun in northern Europe makes me happy, but also there's not currently a lack of Sun in northern Europe Yeah, I love the seasons in Norway too Okay, new question. It's a PhD worth considering if you don't imagine Remaining in academia after getting the PhD. Um, I get this question fairly frequently and The the question as with so many things that have to do with the PhD is it depends A PhD like the the document that you get the certification I don't think matters Especially if you're not gonna remain in academia and if you're gonna remain in academia then it's sort of a necessity But out of academia the PhD is the PhD like title is Very few companies know what to do with it. It doesn't Qualify you for any positions. It doesn't really change your pay. It's just sort of there That there are some companies for which that's not true like if you go into more of like a research lab Like Google research or Microsoft research And I know that there are some positions at Google for example that sort of require a PhD So there are exceptions to this but by and large a PhD is Excessive as far as the title goes That said that the value that I got of my out of my PhD was not the title It was the six years that I got to spend Doing things I thought were interesting and learning things and branching out. That's the real value, right? If I had worked for those six years, I'd be sort of constrained to what my employer Wanted me to do or was willing to pay me to do and maybe I do some personal development on the side there But during the PhD, I had a lot of autonomy over, you know, which directions I wanted to pursue That's how I got into Rust is how I got into Tokyo. That's how I got into distributed systems, you know built Noria Got into a lot of the low-level concurrency stuff like so much of the things that how I started streaming How I had time to write the book, right? So much of that came out of being in a PhD program not to mention that you have, you know, very Talented people guiding you through the process. You're working with other people who have similar interest to you and also have similar sort of Freedom so I I really enjoyed the process. I don't think that end result in terms of Certification is worth it. But the process was worth it for me It also depends on what you mean by worth it, right? So if you attend a place where it's covered by a stipend or like a research assistantship or a teaching assistantship Then I think it's at least for me pretty clearly worth it because I enjoy that kind of work If you don't like teaching don't really like research and have to pay out of pocket for it Then no, it's probably not worth it So it's it's hard to really give a an answer that applies to everyone here Is a master's better in that sense that depends what country you take your masters in so What the difference between a masters and a PhD and what they both entail varies a lot by Continent but also by country. So ultimately what was valuable to me was the relatively unconstrained research time and If you get that as part of a master's that sounds great that may be fewer years where you get to do that But it but that might be a worthwhile trade-off for you But for example, there are master's programs that are maybe three or four years But you have to take classes the whole time so you lose a lot of that ability to Budget or spend your time as you wish There's also the opportunity cost like there's a very real opportunity cost even if you don't have to pay out of pocket Then you're also not earning a programmer salary during that time, right? Which is a pretty severe cost And so worth it is very hard to answer It's not the destination it's the journey exactly right What's your PhD and my PhD is in distributed systems and databases, I guess Okay Next question. What is your partner's opinion of Norway? I Would I would have to ask her It's It's weird because we moved here June We got to Norway June 3rd, and we moved into this apartment June 18th 15th 18th somewhere around there and And so there's been a lot of moving like the past two weeks as well have really just been Still moving even though we moved in now a month ago Because it's like you have to get computer parts you have to get furniture furniture has to be delivered Then you realize you need other furniture and it's just like there's an endless cycle of just like Constructing a home even though like we're renting so we didn't have to like buy a kitchen, but but even so you have to fill the kitchen So I think everything is felt very Temporary very in motion very in flux And so I think now it's finally starting to settle down and we actually feel like we live here now And so I think I think her opinion is very Colored by that by the fact that we don't really have a That we haven't really gotten to experience it the The way that it will be when we live here That said I think and also There's a heat wave in Norway or was until very recently So we moved here in the middle of summer. So like the sun was out for like most of the hours of the day Plus there was a heat wave. So it was extremely warm So we moved here and she was like you told me it was cold and dark and it is the opposite of that Like to the point where it's like sweltering and in Norway because you know, we don't really get heat To the same extent as the rest of the world People don't really have air conditioning. That's not a thing here But of course in the US everyone has air conditioning and so we came to a place where it was like it was Swelteringly hot some of these days So we came to a place where it was swelteringly hot and no AC and it was bright all the time So I think that made it a little I don't want to say miserable but but it put things off a little on the wrong foot But I think now we're starting to settle in I think You know, we went to dinner at a restaurant yesterday because like the first time we've done something fairly normal We now have most of the things in the house We're not waiting for any major pieces of furniture anymore except the TV bench, but where we're getting there So I guess the answer is ask me again in like two months and then we'll see Let's see Yeah, I know it's it's also only gonna get darker and darker and colder and colder over the next few months So hence again, ask me in two months and we'll see but but I think you know in some sense I think she's prepared for this like she she knows that that's what Norway is like and we'll we'll just have to see whether that ends up being okay or not and For what it's worth like I don't know if I'm settled on Norway yet either There's a reason I moved away Like I guess 15 years ago now. Wow There's a reason I moved away in the first place and it wasn't just to like explore the world Um, and so it could be that a year from now now my partner's like Yeah, let's stay and I go. Nope. I need to leave that could very much be the case. I genuinely don't know Uh, but at the same time, you know, I really like Norwegian winter I think it's nice. It's pretty it's it's different. It's cozy Um, and the question is whether I can make that rub off Um Okay Next question Um, what's your recommendation for practicing rust? I've read a bunch of books But I still feel like I'm not sure what I'm doing Do you have some interesting project ideas that could touch on some important rust topics? um The best way to practice rust is to build things that you need And I think I've I've stated this a bunch of times But like there's just no substitute for just Building things like building real things and if you're going to be building real things Build things that you use or build things that you need So either find a program or tool that you're already using and you want to make better and start contributing to it Or ideally start fixing You know bugs you found in it features you want in it Or if there's some tool that you feel like is missing then go build that and do it in rust That's the By far the best way to get experience is to do that hands-on and to be driven by something you're interested in If like what's interesting to me is not necessarily going to be interesting to you So it needs to be driven by your interests and when I say something you need There's a there's a slightly variation of that which is something you want to build right So if you're like man, I've really wanted to learn how databases work or how sequel works Then go build the database like the the docs for sequel light are really good. The code base is pretty readable Go build a database Uh, it might not be something that's you're going to use for anything But it'll be really interesting and as long as you feel as though you're learning something You feel like it's interesting You feel like you're you're driven to do more about it then that's great That's also learning but but ultimately you have to build something real The the reading the books is It's a useful step ladder if you will but but it's not going to get you all the way Okay next question AI relies a lot on GPUs. What is your take on rust for GP GPUs ignoring all the web GPU stuff The focus is modern shaders and graphics should the rust line team work on proper hardware accelerator support at the language level um I don't have too much to say on this topic yet. I haven't done a lot of rust gpu programming I've done some gpu programming in c++. Uh, and it was Honestly, it was kind of fine. I wrote some open cl code. Um and I think it is nice when the language is The language you use for programming the gpu is similar to the language that you use otherwise And I think rust could mimic that I don't know that you want just like gpu hardware acceleration for the language that that feels Um Like something that might actually be hard to pull off and have it not be full of leaky abstractions um, I think what we might want is like a GPU version of the rust language that again then gets compiled to gpu's Um, that said, you know, it might be that you can totally do this through just library APIs and make that be pretty nice But I don't know. I don't have a great answer for this. Um, but but I think there's a I think there's a space for although it is a lot of work a space for a sort of Rust like language that targets Open cl and and kuda. I think that would be pretty neat. It might already exist um Okay Next question Where in norway will you be working from do you plan to be active in the local rust community? So I moved to oslo, which is the the capital in norway. It's like in the southeast corner of norway and I'll be working from home. Like this this space right here is also going to be my office space If you remember from my streams in in los angeles, I had two separate desks one for work and one for my You know home computer and streams and stuff Which I really like I like having them separate so that I go into like a physically different location to be at work Or not at work. It just helps mentally a little bit here I don't really have the space to do that So I'm just trying to make this room as nice as it can and then try to find other ways to make it clear What is home and what is work stuff? um But I'll just be using this this one room as my office for all the streams and home projects and stuff as well as work um And yeah, I mean my plan is to become active in the local rust community There hasn't been a rust meetup arranged in oslo for a little bit But the next time one is arranged. I'll definitely be coming. My plan is just Start to come to them regularly um Los angeles, I think used to have rust meetups But then they sort of ended for a while and they were remote for a bit which doesn't interest me as much Um, and also in los angeles everything is far and you have to drive which is really annoying Whereas here Basically wherever it gets arranged I can just either walk or take public transport, which is really nice So it makes it a lot easier to be to be active I might also end up being active in the sort of adjacent rust communities. Um, so the The Copenhagen rust meetup for example, I think I'll probably go to not super regularly, but every now and again Maybe I'll give a talk there. That would be pretty fun My brother also lives in Copenhagen at the moment. So I have a good excuse to come there for for both And same thing with with Gothenburg or Stockholm That sounds like fun to go visit too Helsing also has companies in Berlin, Munich, London and Paris And so chances are I'll be going to those places Somewhat frequently as well. And if there are rust meetups happening at the same time, I'll go there And the other thing that's nice about being in europe now is a lot easier to go to Most of the rust conferences except for rustconf Um Rustconf is more annoying to get to now because especially with a new location There's like the only flight is a two-stop flight Um, but so like rust fest rust lab euro rust I'm sure there are others. Um, I'll I'll be aiming to go to Which I'm I'm really excited for Oh, yeah, isn't my shirt great? I'm very happy with the shirt. It's great Um Follow up question to the office. What chair is that? Oh this chair. Um, what's it called? It's called like the Uh It is called the arasi vernaza Very very fancy name It's a it's pretty comfy. I kind of like it Uh, it is this I put it in chat Um How do you stay productive when working remotely and how to stay disciplined that that's tough like when I work from home I I don't actually it's not a problem when I have a separate physical space because I can just go over there and Basically track my hours. Uh, I I have no problem being remote because that's how I work in open source too It's all remote. It doesn't that doesn't bother me I I don't need to be around people to stay motivated to work. Um, I also, you know, try to choose jobs And this is also a privilege But like I I choose jobs where I enjoy what I do And so staying motivated is usually not that hard because I'm working on interesting problems quite to the contrary I need to make sure I don't work too much because the problems are so interesting. Um But but I am interested in seeing what it's going to be like now where my home space and my workspace are one in the same I I worry that that's going to make it's going to start to blur the lines And it's going to be harder to focus at work because the home things are so close nearby And it's going to be harder to focus on home stuff because the work stuff is so Will so easily distract me. Um, I I generally don't know how that's going to work yet And and I haven't yet figured out what my sort of Uh tools are going to be for for separating the two Um Okay Have you ever thought about moving to vs coder id instead of neovin? No I have basically no interest in having a non-terminal editor Um, there are a bunch of reasons for that. I just I have no There's nothing I miss Or rather, there's nothing I wish I had any of him just at all. I'm very happy with my setup So I I don't really see why I would switch because there's nothing I would switch for Like there has to be something that pulls me to the other thing and there just isn't at the moment um And so You know, I've thought about it, but that there's just nothing pulling me in that direction Uh, and then I need to like have a separate like gooey window open and like one of the things that I do a lot And you'll see this when you watch me do streams and stuff is that I will actually Exit my editor and then run cargo commands because I enjoy having I enjoy being able to customize the commands. I enjoy being able to do it all by the keyboard Um, I enjoy being able to see the full cargo output I don't want it all to just be like buttons and pop-ups in my gooey Uh Yeah, I mean helix too seems like an interesting enough editor, but at the same time I I don't really feel like I have anything missing from neofim at the moment Um, and that might change, you know, like I did switch from vim to neofim Um, and there could be someday in the future where helix is good enough that I want to switch But at the moment not really um Have you tried co-pilot and vs code? No, I haven't tried co-pilot anywhere because I don't I don't really want ai to write code for me. Maybe that's that that's just me being old and grumpy and old school and that it I'm going to eventually get to a point where I see it being useful Um, but at the moment at least there's not that's not really a pull from either But my bottleneck is not my writing speed. Um And and I recognize that like co-pilot can help you things like not have to look up api docs and such and Point well taken, but it's not it's not compelling to me at the moment um And no, I haven't tried helix. Um, I really like them key bindings And so anytime I try a new editor my experience is well, but the key bindings don't quite work like in them And that makes me sad Okay Since you're going to be working for a company in the defense sector Were you able to find out if the company is not involved in evil projects? For example, palantir if yes, then how? um so the the um The mandate of helsing is a little different than um companies like palantir um What one of the things that I Spend a lot of time thinking about before joining helsing was exactly this kind of questions of like How do I know that I'm not? Getting into a space um That I wouldn't be comfortable saying that I work in Right, like I need to be able to live with myself working for this company I need to feel like the work that I'm doing is is not just like valuable or important, but it's also like Good or at least not evil um And ultimately I managed to convince myself that that is the case for helsing at least for the time being um and A lot of there are a lot of complex Essentially mental analytics that go into that decision um As I don't know that I'll be able to fully articulate it but but part of it came from Talking to several of the sort of senior people at the company about How do you make sure that this company isn't evil? How do you think about the fact that that the The bar the line that you might cross is so tricky um and also the recognition that like It's it's impossible to guarantee things in the space like defense and war are unpredictable beasts and so For example, it would be infeasible for a company to build technology and say this only works for defense Because realistically defense is also offense and very often the technology is sort of um Isomorphic like it sort of works in reverse as well um So so the question becomes not so much Which technologies do you build it does to sometimes that there are some technologies that are purely offensive Although you could argue. Well, maybe they could be used for defense too. That's true, but but there's like things like um Uh internal surveillance in a country for example that is Much more borderline for me much more gray that i'm not super comfortable with but also not something helsing does um and I think the calculus for me came down to the fact that helsing Isn't trying to say We would never Build anything that could be used offensively But but rather we try to choose who we work with and try to choose to work with like You know european democracies to meet a uh high bar for moral decision-making democracy Responsible use of um technology and defense and that's not perfect either but But realistically, I think that's sort of as close as you can get I think you have to choose your partners. You have to choose Who you are Giving the capabilities you're developing to I think it is a losing battle to try to say I only build things that work defensively um The link to helsing is helsing.ai That excludes the uk we sell far too much to the saudis. Well, what's interesting is also that This isn't weapons expert export like helsing doesn't build Hardware they build software and so if if helsing builds something for The uk military for example, it's not something that the military can then sell on to another Country at least not that that's not really how it works um I don't have the insight into the company for like how lending Equipment and such works for example. I can't really speak to that um, but the the The goal here is very much to try to choose responsible partners um You said you won't use ai but the company screams ai. What do you do over there? um Yeah, so the company does a lot of ai, but I'm not doing the ai parts I'm not doing like model tuning or working in pie torch or anything like it Quite to the contrary. I'm working on the rust stack and on expanding the rust stack And I'm finding ways to take the stuff that's currently python and moving it more into rust where it makes sense Um, and so this is much more like rust infrastructure or distributed systems Then it is ai my role that is at the company. So the company One of the big things that does is is sort of ai enabled technology But there's a lot of technology that surrounds and underpins the ai parts and it's that technology that I work on um And yeah, I mean defense does have a reputation to be a bit slow tech wise that's certainly true And I think that's one of the things that's really interesting about helsing is that they're essentially trying to change that game right they're trying to say The the current way of thinking about Software in defense is broken. It does not work and if you look at You know nations across the world that we're more concerned about um They're not making the same mistake. They're making heavy investments in Software over hardware and we need to do the same if you look at like the war in Ukraine too One of the things that the ukrainians have had a lot of success with is using technology It's not by building fancier hardware. If anything it's taking less fancy hardware and doing smarter things with it And the way that you do that is you lean into the software engineering of everything you you build Ways to like do rapid deploys software engineering at scale feature improvement like things don't take years they take weeks That kind of timescale and then the question becomes well, how do you square that with moving deliberately and And without making potentially extremely costly mistakes. So it's not it's not obvious um Okay Next question Again, if you have questions in chat, please put them in the The q&a page because chat disappears so quickly and it's it's hard to It's it's hard to keep up with it on the side So post it there and then other people can vote on the questions too. Um, if you are In here and you're just curious about questions Go to the q&a page and like scroll to the bottom and look for new questions People have asked that you think are also interesting and vote them up So that way we actually get questions bubbling to the top that are interesting that are not just the ones that have been there The longest so scroll down the page a little bit and look for questions that you might want to vote for How are your cats the cats are good the cats Have been on quite the journey. I mean we brought them with us from los angeles to here And so the first part of that was making sure that they had all the like paperwork they needed Which was a whole logistics nightmare of bureaucracy everywhere And the only thing they really needed was a rabies vaccine, which they've already had but we needed another one to like make sure the paperwork was right But then there was just so much paperwork that needs to be signed and counter signed to be brought in the scent in the mail And it has to be like physically embossed pieces of paper that you have to bring with you and it's a whole thing Um And then you know, we gave them some sedatives before the flight and then we put them each in their own Little soft cage We brought them on the plane in the cabin and then they go under the seat in front of us So they were in the cabin with us and It worked pretty well They were a bit restless in the days leading up to the move because we were just packing everything down to the house was just There was no furniture everything was just disappearing around them which freaked them out And then we brought them to the the airport and that was pretty stressful because there's lots of people lots of sounds lots of smells But then the the sort of sedatives were starting to to work and so they sort of managed that decently well and we covered their little Carriers and blankets to pad the sound and give them some darkness and feel like they were in a safe space And then you know the flight is very long the flight was 12 hours from lax to Copenhagen And then we had two hours in Copenhagen like a stop like a layover And then we had a 50 minute flight from Copenhagen to Oslo And I think that Copenhagen to Oslo flight was actually the worst one because by then the sedatives have worn off So the cats were like pretty unhappy And we were really tired and that plane wasn't as nice right the the long distance planes are generally The long-haul flights are better than the short-haul flights And so that was just a lot that that last flight was pretty rough But you know we we got them all the way through and without any incident like You know we had a lot of pee pads in their little carriers and stuff. They didn't need those You know, they didn't really eat or drink because they were nervous but also sleepy So they just sort of lay in there and we would you know pet them and soothe them whenever they woke up and got Worried about where they were and what was going on And then when we got to oslo We then have to take them on a train from the airport and into the city And then a car from where we got to the house As we were staying at my my dad's old place just while we were waiting for the apartment And so then they got brought to this completely new house and let out there and then you know They had to explore that space and over the next two weeks. They got used to his house Uh, but then we had to pick them up and move them again to here and this was just an empty apartment So now everything was stressful again But now finally they're starting to settle up. They found the little spaces They've started playing again, which is the thing that they hadn't really done since A few weeks before we before we moved from la because they'd been so stressed So now they're finally in like a playing mood again. They're starting to get cuddly again So they're they're getting better. They're like picking their favorite window cells to sit on I don't currently none of them are in here. I'm trying to get them to I'm trying to get them to make my office their base, but we'll see whether I succeed Uh, let's see. Oh shows the cats Miso Miso Miso do you want to come here? No, I think they're sleeping. It's the sleeping time of the day They sleep until usually around like 8 30 And then they want to play and eat So they're currently all sleepy Yeah, chai is called chai because her coloring looks like milk in black tea And miso is named miso because I don't know. I like miso, but also because she's like white yellow and black I don't think you can get black miso, but you can get white miso and yellow miso So that's close enough. It's also just a cute name Let's see How do you navigate the stress of relocating and interviewing at the same time that was pretty rough Not gonna lie that because especially because The interviewing was happening like The same week that all the packing was happening and we were packing everything down into boxes But also we couldn't pack everything because I needed things for interviewing like my computer setup and stuff And so progressively my interviews were done with worse and worse hardware all the way down to like The laptop on my lap And so that was stressful and I think it was stressful for my partner as well because You know it meant that I wasn't around to help with as much of the logistics and packing Because I had to be you know emailing people and looking at doing interviews and following up on LinkedIn requests and whatnot. So that was that was a lot of time Which was unfortunate like I wish in some sense. I wish I'd started that earlier The the reason I didn't start it earlier was because it didn't really It didn't really work with my departure from aws, which I also I didn't know that I was leaving until a little bit later. So it was hard Oh miso paste can be dark or black. It's true. Yeah, Misa's great Okay, next question What is your daily routine? Um, let's see So My daily routine is about this. I'm expecting it to be about the same here eventually as it was in LA Which is I usually get up around like eight eight thirty And then I aim to start work around nine ish I'm normally more of an evening person but For whatever reason I've started waking up earlier and I don't love it Maybe I'll I'll shift my whole day. This is thing called the It's a Danish thing the B people society Yeah, the B society right, so um Here let me send you this. It's great So There are a people and B people a people a type people get up early B type people get up late And I'm definitely a B type person, but the the whole observation from the society is that the Our entire society especially in western culture is based around eight type people schedules So like people expect you to be at work at eight or nine People expect to have dinners at certain time and nothing is open late at night You know if you have kids and bring them to kindergarten like that closes on like a people's schedules And so this observation was like that's unfair to like the significant portion of the population which are B people They just don't align with that so they live their lives sort of offset from their optimal routine And their claim at least this was their claim a few years ago was that Honestly, the way to do this is we have to split society We have to like have a people work and B people work We have to have a people of kindergartens and B people kindergartens a people schools B people schools um And it's extremely costly and probably never going to happen But but that's one of the reasons why I've ended up becoming earlier in addition to I think the sun and getting older Is that um, I think I've now just Accepted the fact that society operates on an earlier schedule And so I'm at a disadvantage if I don't follow the same schedule and it sucks I I like sitting up like late at night when it's dark and quiet. It's great Morning's not a big fan. Um, but I've sort of ended up there So I get up around 8 8 30 I have breakfast, uh, which is usually just Uh granola, muesli yogurt and blueberries or strawberries Uh and some honey Very good mix. Uh, this is breakfast and then I work from nine ish until Like 5 30 ish. I usually take about an hour for lunch somewhere in there. Um I try to make the way that I try to work is I work when I'm productive And if I feel like I'm just not productive right now Then I go do something else and then I make up the time elsewhere Um, because there's just no point in me sitting at a desk not being productive That doesn't help anyone. It doesn't help the company I work for Um, arguably I shouldn't even need to make up that time, but that depends on the company How forgiving they are with that. I don't know how Helsing is here yet, but um But but I'm certainly of the opinion that Requiring people to sit at their desk if they're at that time not productive as long as they're generally getting their work done Is probably counter productive. Um, and so I try to identify when I'm sitting there and just Slogging through and at that point take a break and do something else go play with the cats go outside You know touch grass Um so I do that lunch I bake sourdough because that's the thing that I do now. I'm that kind of person apparently Um, I'm very sad about my sourdough here. I didn't have a chance to bring my latest batch So I had to use an earlier one. I've dried and it's just not as good. It doesn't rise as well I need to like revivify my my sourdough yeast. I tried to bake it the other day and did not work well Um, so I have some sourdough for lunch with like usually like avocado Which is harder to get here. Um, and caviar cheese because caviar is tasty and cheap in Norway. Um And then work more and then end around 5 30 and then, um Um Dinner maybe around 6 30 or something usually we make it at home um And then Play with the cats and feed them. I do that in the morning and do that in the evening And you know then Whatever I want to do in the evening really watch tv Open source stuff and then usually on weekends You know, that's sort of what I consider my free time But of course or do streams is all the things I do in the evenings Uh, and then during the weekends, I do things that I want to do Which often includes just like doing things with my partner open source stuff My personal email which takes a lot of time So much time just goes to like Open source things like things that I've given my time away to Um And then Monday is back to work again Well, one of the things that I really liked about my position at AWS and I think I'll probably end up doing something similar at Helsing is to have my educational rust streams be Essentially a part of my work So I would take, you know a friday every few weeks to do Or a half a friday or something to do a crust of rust stream or some other rust stream And that would just be It wouldn't be on behalf of amazon like amazon wasn't sponsoring them There's no amazon branding or anything that they know how to say but it was more like I was allowed to take personal time to do the streams during work hours kind of arrangement And and that worked really well and and I you know after having spoken to a bunch of the people at Helsing I think I'll be able to do something similar there after a little bit of time I just need to get oriented a little bit first and some Contract things need to be figured out But but that really helps with the daily routine because If if I can't do streams on work time It means I have to do them on weekends Which then steals away basically a day of my weekend whenever I do them, which isn't ideal Um, and as I mean especially for the longer streams like if I do a seven hour stream, that's my saturday gone, right? um Oh the a person b person concept is not specific to Denmark But the b society I think started in Denmark Uh, let's see Okay Next question Can you talk a bit about the state of the amazon build system and rust what kind of impact on its usage of the company? Did you feel you personally had? Personally had is hard to answer. Um but I so Amazon has this Internal build system called brazil think buck basal is a similar kind of ordeal. It's a little different But you get the idea is like an overarching larger build system and then you write sort of integrations between That larger build system and how you want to build any given package And then we have sort of standard integrations for many of the supported languages like rust And I built the integration between brazil and rust or rather One existed that was community built before I joined amazon and that was pretty good It was in a better state than many of the other uh language integrations But one of the big things that I did was I rewrote that integration to be much more cargo centric so it was focused on You can just use cargo the way you're used to outside of the company Inside of the company So when you were using rusted amazon with this new system What you would write as your sort of build system rules is really just a ci script that just runs a bunch of cargo commands And then there's a cargo sub command called cargo brazil That will then Deal with anything that needs to go across the boundary to brazil things things like pulling dependencies from brazil into your cargo build or publishing things out of your cargo build into brazil as Sort of output artifacts things like binaries shared libraries that sort of stuff um And so I built that integration and I think by the time I left You know, it was About 50 percent of usage had moved Without any really forcing function to make people move is more like announcing that you can do this now And then people got pretty excited and started moving And I think that made the experience a lot better I did a decent amount of work on security integrations Which I can't talk too much about the details there, but on security integrations of the build system and stuff As well as things like provenance tracking and auditability of dependencies that kind of That kind of stuff And then I did a lot of just sort of guidance. So other teams that were building things in rust Whether in service teams at amazon or in other build teams Just helping them with making decisions solving the rust problems debugging the rust builds, um, you know Dealing with future proofing or upcoming breaking changes in rust that might break them like the um When rust changed which version of the android ndk they built against that had some Some repercussions internally that I dealt with so there's a lot of really just sort of glue like holding together all the different parts of the internal rust ecosystem Which was really fun, but also, you know pretty exhausting mostly because I think we were a team that Needed a lot of people and and I think this applies to every language ecosystem internally like if you're going to own the language experience That's a pretty hefty task if you want to do it properly. Uh, and I I think our team ended up in a place where we didn't quite have the um the amount of investment and Um Support from our leadership that we think we needed to to to truly excel at the task that we've been given um Oh, yeah, the the name brazil comes from brazil The the amazon river going through brazil um Okay, next question How often do you exercise and does it help you write better code? I don't really exercise and I should and it would probably make me write better code um Realistically the answer here is I I used to rock climb a decent amount now. I haven't done it in ages and it makes me sad. I want to get back to that Um, hopefully that'll be easier here because I can actually walk to places rather than having to drive there again I hate having to drive places and especially when it ends up being far um But also I I started um jogging back in LA and I started jogging like almost every day Um, which is maybe an aggressive place to start But I started jogging maybe A month or so before we moved and so that of course then got completely interrupted by the move Um, so I'm hoping I'll get back to that here Um, but I haven't picked that back up Um, it does mean that like my body is pretty tight So like I need to like stretch and stuff and get um massages and things to like Help with the fact that I sit a lot. Um And that's unfortunate and you know exercise is one of the things that that helps with that as well Okay, uh Next question What's your take on the whole AI replacing engineers hype and if you believe in it Do you think there are any areas in the software industry apart from ARML that might be insulated from it? I don't think AI is replacing engineers. Um, at least not for quite some time I think what AI does is it changes the way that you build software In some cases in other cases It changes the the role of the engineer to to being more of a debugger and Guider than it is a programmer at the same time AI just isn't good enough yet to write code one of the problems and I've had this conversation with a bunch of people now that At least my perception here is that AI the AI is really good at pretending to get things right um And it does really well in spaces where the delta between almost correct and correct Is imperceptible to humans or close to it The problem with engineering is that the delta between almost correct and correct is like a chasm If you have an almost correct program Then it is wrong And if it crashes, then that's great. But if it doesn't crash, you're in a really tricky spot And I I think that the problem we're going to run into or the problem you run into when you try to use AI to say Write microsoft word is that it's just not capable of that level of Systems design writing that level of code. That's all correct Engineers aren't perfect either that that is totally true. But but I think there's a there's a There's like an order of magnitude difference there somewhere That's not to say that AI won't get there But i'm not really seeing it replace engineers at the moment And I think it'll take some time to get it there because getting from 99 right to 100 right is pretty far And that that correctness like there's just very little room for error At the same time, it's a constrained domain like you have a compiler that can help you more strict languages might help here might let the AI like iterate by checking things and Proposing something and it doesn't compile and trying again But but I think realistically maybe over time we'll see engineers move more into a Debugging guiding qa kind of role But I'm not worried about that happening anytime particularly soon But it's hard to say it's it's you know an area to speculate in I also think that there's going to be A need for engineers to Do like the larger systems thinking for quite some time Like I feel like a lot of the engineering I do is like Gluing together different systems that have very little documentation No one knows how they really work and it's partially like figuring out how to how they work. It's pretty interactive Like not quite reverse engineering, but there's a lot of like trial and error In a way where it's not as though There are just facts from elsewhere that get injected into the program and then it's ready I think engineering that does work like that like if you're doing Um web design for example That there you might get into more trouble because the precision isn't quite as necessary there as well as It's I don't want to say formulaic, but it is More of the stuff that goes into the output there are examples of elsewhere that the AI can have learned from Um, which is less the case in like low-level stuff interacting with hardware that that area Um LBMs are amazing predictive text engines Yeah, I mean they're they're very good at generating realistic sounding text But that's different from generating correctly syntaxed and semantically correct code um Oh, yeah, not to mention Making yeah, someone pointed out in chat making changes or extending AI generated code as a nightmare And I believe that because one of the things that as engineers we get really good at is not generating code for a given task But figuring out how do we write this code in such a way that it works with all the other stuff? It's not writing code in isolation um Let's see Generating things like tests is a great use of current ML tools. I believe that Um, and that's partially because that space is is fuzzy. It's not as though you're targeting something Correct. You're targeting a bunch of different things where many things are a correct test You might like really realistically there are almost an infinite number of tests, right? Software engineers will bog towards more formal specification and let the AI extend compilers Handed automated reasoning code genin simulation. Yeah, that's the other potential path forward here right is that engineers end up doing a lot of the Specification of what does correct even mean in the space? Maybe AI does code generation or even proof generation Guided by humans But ultimately like rewrite the definition of what they're the inputs What are the outputs and what does correct mean and then the AI just sort of generates everything that goes below and then you have like a You know use formal verification methods to check that what it generates actually matches the specification That could also be a path. I think that's even further along Getting the AI to generate Both getting good enough proof assistance and getting the AI to generate reasonable proofs And in such a way that even if it gets the proof almost right, you can get it to be right And getting it to generate the code that matches the the proofs like I think that's a long long journey So again, not particularly worried about that Um I'm not sure I trust an lm to generate tests that cover all the edge cases I care about I don't think that's the right way to think about it. I don't think you want the lm to generate all of your tests I'm saying that you could add this to your testing arsenal So in addition to writing the tests that you know, you have to write in addition to things like, you know, you might use tools like Prop tests or quick check or something like it that does property-based testing or fuss testing Those are valuable too, but they just randomly walk the space um And those are also valuable and you could imagine that AI becomes another tool like this where you describe I have a function that takes these kinds of inputs and produce these type of outputs Generate some challenging test cases for this and it might generate something that's somewhat reasonable and now you have more test cases So I think that's a totally reasonable way to try to um try to do this or try to use it as a tool Um Have you seen some of the recent developments with 1 billion token transformers? That would mean the AI could have exponentially more context than it currently lacks One of the challenges is that the context here isn't Context that's available at training time It is context of the problem space that i'm giving it a prompt in right like the context is in the prompt It's not in the training data and that's where you run into challenges um Like I need to tell it about all of the code in which I am trying to write this function for example We need to turn coding into math. I mean that's what formal methods do um It's good to write and test before you write your code But if you provide it with code and say test this nine out of ten times Write tests that will pass 100 of the time um Yeah, I believe that so it's essentially like test-driven development, right? Um, and I I totally believe that it works better in that kind of setting Uh, all right. Let me I'll be right back. Let me see what this phone call is I'm back It was a delivery I was waiting for so I needed to pick up the phone so they don't just leave um All right next question I assume you can now see me and hear me I guess one way to find out is to go look at this Yes, fantastic. Okay, uh mark has answered Let's see next question the job offers of helsing aren't remote only Are remote aren't remote only as far as I can see will you work in berlin munich paris or london at least part time? um so This one's a little tricky so Because of the domain that helsing works in there are a bunch of things that just have to happen in person Or have to happen on location and so as a result helsing is at least in part a sort of A company that is pretty focused on location Location based work, but that said there's a lot of work that has to happen, you know across countries across locations um and so And and it's not as though you can't work from home. It's not as though you can't Work remotely. It's more that it's a little trickier than for companies that are in different In different areas and so as a result Sorry don't panic don't panic all right Let's see so the Yeah, so so the the positions that they have are all the ones that are listed are at location because many of them require That you are near a location so that for the work that requires you to be in a particular location You're actually near that location But in general what I would recommend is that if you are Interested in any of the jobs that helsing has you should get in touch anyway Um, even if you're not near one of those locations and we can see what we might be able to figure out um But but the company is like there are some things that will require you to be on location In my case what will probably happen is that I will I'll be working remotely from here But I will probably be traveling to the other offices Not necessarily frequently, but with some regularity And part of that is because the the work that I'll be doing Isn't going to be tied to a specific location all that often I won't be working on a team that like works, you know, specifically with The german military right instead. I'll be working on more infrastructure distributed systems kind of stuff And so as a result, I'll be working with many different teams that are in turn working on location Uh, and what that means is I have more flexibility both to work remotely But also to work across these locations Which is all to say if you're interested in this kind of work get in touch and there might be similar kinds of roles Um with similar kind of flexibility that might fit you. Um, same thing with like if you near live near some of these these places, um, then You know, it's not like you have to come into the office every day And so they're pretty flexible in terms of arranging for you to work from home, for example Okay, um Favorite purchase of the last six months Of the last six months Oh, that's tough um I was gonna say like I've bought a bunch of furniture, but I don't know if any of them would be my favorite purchase Okay, I'm gonna give a slightly boring answer, which is When we traveled from LA to Oslo, I decided to buy Not coached like not economy tickets, but buy like the plus class not they don't allow cats in business Otherwise, I would have bought business tickets, but I bought like the economy plus kind of thing and that I think was one of my Favorite decisions to have done because trying to travel with two cats And the two of us and we had 10 checked bags Trying to do all of that in like In coach, I think would have sucked Uh, and so I'm very happy with that purchase, but it's not a purchase of a thing. It's a purchase of like luxury Upgrade, um, but but I think I think it qualifies um Let's see Is reducing your job pool to remote only something that made you miss out on great opportunities elsewhere That's very hard to say right like I don't have the counterfactual here. I don't know whether um I don't know whether There are interesting job opportunities that I did not get because I announced that I would only work remotely But at the same time I don't really care Right like for me. It is so important to be able to choose where I live and to be able to Work from home and to be able to like Be around my girlfriend be around my cats have lunch at home like that all matters so much to me That if there was a really interesting job that requires me to go into the office all the time I that trade-off isn't worth it to me um which is all to say like It may very well be that I've missed great opportunities elsewhere But I'm okay with that if if that is indeed the trade-off that I had to make then that's okay um Okay, next question in the long term. Do you plan on being an individual contributor throughout your career? Yeah, I mean I have no desire really to go into the managerial track I think You know as you become more of a senior engineer you end up taking on more Sort of technical leadership tasks. So they're not necessarily management, but they're more like Strategy direction Mentoring which maybe sounds kind of managerially, but it feels different um But yeah, I'm not really planning to go away from an ic role. The closest would be if I end up in a sort of cto kind of role then maybe that would be the closest um alternatively, you know, I've been thinking of going into sort of consultancy and essentially do Rust consultancy and That would also be a little different for me an individual contributor, right? It would be more sort of come into companies where My where I might be able to make a difference for them, but that's not necessarily writing code for them that might be, you know coming in and helping them with Strategy decisions about how to adopt rust it might be coming in to do teaching which is also not really an ic kind of Job function if you will Which is all to say like I really like writing code and I think I will be an individual contributor For the rest of my life in some capacity, but it might be mixed with some other things like like education, for example um Okay Any reasons for leaving amazon other than better opportunities and location? Um So my departure for amazon is is a little bit complicated, but but not super complicated essentially It was a combination of I didn't feel like they I feel like they were leaning in they were leaning away from remote work and I think the world is moving towards remote work and I certainly am and so the company is moving the opposite direction of what I am And that just inevitably leads to friction down the line In addition, I was expecting that the compensation they would offer me in europe would just not be compelling And part of the reason for that was I went through a promotion cycle where I did not get promoted to principal and basically got told Try again next time and then you'll probably succeed Which is a little demotivating right like it doesn't it feels weird But that also correlated with me not getting the compensation with the move that I was sort of hoping for So so those are two of the parts and then the third part is that I think the company wasn't making the strategic investments in the Sort of build system space for rust that I was hoping they would Or rather, you know, they started making that investment and then sort of tried to hedge their bets And and I I felt that like we didn't quite get enough people that we wanted and also they were hesitant about the people that we didn't manage to get not in terms of the quality of them But just like are we spending too much resources on rust should be spending more on other languages that are used internally and and that that made it it made it um It's a little bit of a frustrating environment to work in where you feel like Your Your leadership chain is sort of isn't all in on what you think is really important No matter how much you're articulating the value and the importance of it And so that was pretty frustrating and so all together, you know, I sort of felt as though I did really good work at amazon. I feel like it was interesting. I feel like it was important I think I enabled a lot of cool projects as well internally Um, but I also felt like it was Time with the move to be like, okay. I'm gonna go elsewhere and see what else there is to do Me so come here Do you want to say hi? Hi me so come here Yeah, come here Well, you're gonna be difficult come here Hi Did you wake up? Did you wake and you didn't find anyone? Yeah, did you wake up and you didn't find anyone? Hi You're very cute Do you want to go say hi to the camera look Oh in there In there a lot of people sitting right in there. I know you really want to get I get it. Okay. Okay. Okay. I'll put you down here There you go Okay, bye um Primogen is in chat nice um Okay, next question What is your take on rust for data engineering tools? Yes I think rust is great for data engineering tools I don't that that's my take. I don't think it's a particularly hot take. I think it's just right um, I think you know people tend I think what the questionnaire is asking is like Well, for a lot of this data exploration stuff Python is great because you can iterate quickly you can get to something working quickly That's all true if if you're just building a throwaway tool If it's something you're just going to iterate on for a little bit and then you have something that works And then you're going to run it once and throw it away. That's fine The problem is it's very rarely true In practice, you're going to iterate a bunch in the beginning and then it's really nice to have python or something That's actually interactive But then You're going to end up using it for longer than you thought You're going to come back and visit it later And then have to understand what the code is doing and why You're going to have to refactor it later other people might try to contribute to it later You might have to integrate with more tools over time The logic is just going to keep growing and getting more convoluted. You need to extend it over time and at that point It's it's no longer a throwaway prototype It is an actual like piece of software engineering and then it is nice to write it in a proper language that And proper language is perhaps too strong here But writing a language that lends itself to doing things Well and not in a throwaway fashion and I think that that is more true of rust than it is on python And so even though the iteration speed for your initial prototypes might be slower and rust I think long term it is often the right decision and then you need to use some amount of Discretion to figure out Sometimes python is the right answer Um Yeah, or you end up throwing it together and then it runs longer than you think it ends up being too slow Or it ends up having to scale to more machines or it has to scale the larger data sets since you need to rewrite it anyway um and like essentially, I think you just Just do it in rust from the beginning Like it's just very rare that you write true throwaway one time prototypes that don't matter um Certainly not in a corporate setting Uh, are you available for mentorship? Um So the answer to this is no Um, and the reason why is I just do not have the time Um, the slightly more nuanced answer is that I do mentorship at the companies I work for so at amazon I had a few mentees. Um, some of them i'm i'm going to follow up with after You know once i've settled into my new role as well Um, and i'm guessing i'm going to end up with some uh, some mentees at helsing too, but but realistically I just don't have enough spare time to devote to mentorship I wish I did but I think i'm more impactful at sort of larger scale one to many communication like this um that i am at Mentorship even though I really enjoy mentorship and I think it's important. Um uh, it's just Not where I end up spending my time at the moment Um How do you organize your to dos? You're incredibly successful some tips and tricks would help. Um Well, I'm terrible at organizing my to dos. So Here's the here's the rough progression um I started using some kind of to-do list app microsoft to-do or something and I Completely failed to keep up with it and forgot everything that was in there forgot to add things there It was a mess And then I started using phone alarms so On my phone I set an alarm for some time in the future um Like not a calendar event But an actual like alarm and I set it for like one of the days of the week Repeat on that day at this time when and I pick a time and day where I think I might be able to accomplish that task And then in the description of the alarm I put what I'm supposed to do what my to-do item is and That works surprisingly well because it will actually make you do those things or at least alert you to the fact that you should be doing them But it also doesn't scale to very many tasks Or to particularly complex tasks or anything where They're not really a thing that you want an alarm for every week um Like I've had some alarms that are recurring for like a year every week But they were like things i'm not going to do for a few years And that clearly doesn't work um, and then I started having a uh in my Shell script that runs when you open a shell I would print out in different colors my to-do list ordered by sort of rank of importance And I wrote it so that the higher the so things would be displayed randomly whether an item would be displayed or not would be random But the more important it is the higher A likelihood there would be that it would be displayed So high priority items would be displayed all of the time low priority items would be displayed like One in 20 shells I open or something uh, that worked Okay, but it doesn't really synchronize across devices It's annoying if I think of something to add to my to-do list on my phone But then I want it to be on my computer for like persistence. It doesn't really work across computers very well I don't really want to check it into my public git configs um so that was decent, but didn't really work super well Um, then I started just having a giant to-do dot md file in my home directory That my shell would just print out every time I started my shell And so it got to the point where when I start a shell I got a screen full of text before I got my prompt Clearly also doesn't really work. It's just worse than the other one um Now I use um obsidian and it's pretty good, but it's It doesn't really work as a to-do list because It doesn't remind me of anything, but it does sync Um, so so I think where I've now landed is I use my alarm technique for anything that warrants an alarm and anything else goes into obsidian And I still haven't figured out how to remember to check obsidian But that is a second-order problem But the alarm trick works honestly really well um Okay Next question Which aspect of norway in general if any disappointed or annoyed you upon your return after so many years abroad? Honestly the lack of amazon Having amazon is just so useful like I've had to now just find so many stores that sell the things I care about and a bunch of the things aren't sold here Uh, and or I have to find some obscure site that sells it. So now I suddenly have accounts on 30 different online stores to buy all the sorts of different things And just the ability to order it all on one site have my order history all in one place It is next they delivery for everything. It is just so convenient um second thing is um It's frustrating that there aren't really it's I have an inclination for whenever I want to buy something I try to find the best version of that thing. So I do a decent amount of research and digging into What is the best x? Uh before I buy ax Um, and that works really well in the us because the us has so many people So there are always people reviewing things. You have stuff like wire cutter, which is fantastic. Um Norway's a much smaller market, but even so europe just doesn't really have the same kind of Review places that do like in-depth reviews like tom's hardware Uh or pc gamer or pc world or wire cutter or spruce or Good housekeeping like there's so many of these in the us and europe doesn't really have them And I think partially that's because there's so many languages spoken throughout europe. So they're not You don't get that effect of having lots of people speaking english So there's an english speaking site that does it for europe You get it a little bit for the uk only But different products are available in different subsets of europe and so that that's a little bit frustrating um, and I think the third thing is um Deliveries here are a little bit of a mess still um, and I think they are in the us too to some extent, but like The fact that in general I have to be home if I want something delivered home And the they don't have like keys to get into The apartment building or anything. There's no code. It's just like you need to have the key and only a few people have the key um, there are a bunch of different delivery companies and Some of them are just really bad um, so It's a little the delivery experience is a little frustrating like mail wise. Um, But that might also be because we now live in an apartment building where there's no code on the door Uh, okay Let's see Uh, do you think it's due to the social aspect or even more of a less consumerism area now? Um, maybe But like people have to buy stuff in europe as well And you know, it's not it's not as though we don't buy things Maybe it's more of a It doesn't matter. Everything is kind of the same, but that's just blatantly false, right? Like I if I want to buy One of the things that I'm now going to buy is a nas like a network accessible storage thing and The ones that are available here aren't quite the same ones that are available in the u.s So I can look up reviews in the u.s But not all of the candidates that are going to be available here um, and I'm struggling to find good like reviews of Nasses that are available in europe, uh, I've now found one. So I don't need lots of recommendations, but like it's just it's um Like what do other europeans do in this case? Like where do you go to find which one is the good one? Because if you just go to a random electronic store and buy the first one you see It'll probably work, but like I want to put the bar a little higher than that um That there are drop-off boxes like delivery boxes, but that doesn't really help for larger deliveries I have a post office that's really close to here, which is very handy So I can get things delivered there and then go pick them up, but they have limited opening hours There's like a drop-off box somewhere near there. That is a little bit better But for anything that's larger that doesn't really work like if I order You know cat sand for example, that's pretty heavy. I don't really want to carry a giant box of cat sand From the post office to home Um, I wanted to be dropped off outside of my door like amazon would and I don't really get that here Um, yeah the q&a is going to be available on youtube later um Yeah amazon is available in Some parts of europe, but amazon europe is also fairly different from amazon us. It's not as Unified in a sense the selection isn't quite as good The prices aren't quite as good the delivery options aren't quite as good They're also not in norway at the moment, which is partially because norway isn't even in the eu Which means we're sufficiently different And it's partially I think because of the high sales tax And it's partially because It's a very small market the incentive to go here is is smaller, but I'm hoping it'll eventually come here and that would be handy Okay, um, I don't see any streaming libraries for rust Is it difficult to implement one with all combinators and rust where there's no garbage collector referring to streaming libraries like aca streams fs2 zio streams I don't know anything about this this question. Um I think they're I don't know what those types of streams are But there are certainly Libraries out there for doing data streams in rust um I don't know whether they're Anything like the streams that you talk about, but I'm going to skip this question because I don't have anything meaningful to say Did you ever think of starting your own company I already have Ready set is a company that I co-founded which is basically the company to commercialize noria So nor is my the database I built for my phd and that That research project was then essentially fostered into a company called ready set Which now build that as a commercial product that I don't think they sell a product you can buy at the moment um Or the reason I say I think is because during my time in the us I basically was not allowed to interact with the company because For visa reasons i'm not allowed to have a second employer which Interacting with the company would count as in an advisory capacity and my amazon contract wouldn't let me so I was like doubly barred um So hopefully now in europe, um, I'll be more able to interface with the company and sort of help them along um But it's pretty cool. They have some really talented people working on it And i'm pretty excited to see it become like noria become a real thing And just this question more broadly of starting my own company um I think if I start a company it will probably be around sort of rust consulting work And I don't know exactly what nature that's going to take yet, but i'm not going to start like a big company that that seems unlikely um Okay What do you think about all of the rustconf scandal? I mean this this topic has been debated to death on the interwebs, but Uh, I think it was Pretty appalling the so The way in which and the decision to uh retract this the keynote speakership I think is Not okay. I think it's really bad. It sucks and not just it sucks, but it's Offensive is the wrong word, but it's pretty close. Um I think it is real unfortunate That the way it happened was through sort of shadowy behind-the-scenes conversations um at the same time I don't think it is as nefarious or malicious as It originally came out as and I think that's been backed up by what's come out since um, you know, I think To an extent this is Humans are Complicated and to an extent it is communication is hard And to an extent it is rust governance Was a mess. Let's hope that the the sort of reform rfc is it makes that better And those things together Ended up in this shitty outcome um so You know, I I don't hold this as like rustconf is doomed or everyone who runs rustconf is bad Like I don't think that's the way to look at this um I do think that it was a An awful outcome, but I think everyone involved recognizes that that is true It shouldn't have been necessary for it to get to this point For people to recognize like this should have been caught earlier and not have happened in the first place But now that it has I think there's a broad Acknowledgement and agreement that things went wrong and that there are The things need to get better need to be improved um Yeah, so so chat is basically agreeing with me with other words which is humans are flawed Communication channels broke down and it's the consequences of unstable leadership. I totally agree um, but but I don't want that to overshadow the fact that This was a Horrible outcome of those three factors like those things are all all three of those things are true But the outcome was still horrible. Um, and you know, I want us to do not just to try to rectify this which is Probably impossible. Um, but to do better in the future. Like it's important that things like this don't happen and that we have um Mechanisms and people in place to ensure it doesn't But yes, the point is it can be bad without being malicious and I think that's the case here um Do you plan on a community discord or something along those lines? This is a good question. Um And I'll answer it a little bit more broadly. So I um I've spoken about this in the past too that I Really want now that I moved away from the u.s. And I'm under fewer sort of restrictions um, I really want to start a little Like Github sponsorships patreon Something I don't know exactly what format is going to take. Um So one of the reasons why there it doesn't exist yet is because I haven't decided What the structure should be what platform it should be if there should be more than one and what it should be for like You know, it'd be pretty easy to just start a patreon Hook it up to a discord and just say Give me money Uh, but that's not really what I want to do and there are a bunch of things that I don't really love about patreon Um, I don't know if I like that model for myself or for other people um So I've been doing a lot of thinking about how I think it should work um the part of the problem is what I really want would require me to build my own thing and I don't really have the energy or time or desire really to build my own thing. Um, but it might be what's necessary um, what I've thought about is in the interim to um Essentially Open up a bunch of different channels for people to support me. So essentially something like uh, Github sponsorships patreon maybe nebula um And something like libra pay or or kofi or something um, and then build a website that will Hook into all those websites and Tell you How much have you contributed to me in total and then give rewards based on that and one of the reasons I want to do this is because the different platforms offer different, um Payment structures for example, so for patreon. I think it's something like every dollar you give someone they get about 80 cents Which is pretty severe But like I want people to recognize that there are differences between the platforms I want it to be easy to move between the platforms if you're not happy with one Or if I'm not happy with one and I also wanted to make sure that If you contribute over time even if you contribute a small amount If you've been with me for a long time I there's some recognition of that too. It's not just how much cash are you giving per month that that feels too crude But but again, like I just have a lot of loose thoughts about how I want this to work I've also thought about things like, you know nebula. I think is an interesting idea of saying All my videos would still be on youtube But maybe there are additional videos or deeper in depth common commentary or Deeper dives or something that goes there and then you're like paying per month But you're also getting other things that are not just me I don't know. Um, I've also thought about something like offering like a rust course That would be really interesting or do like a narrative version of rust for a station That'd be really interesting. Um, but that would need to be some other forum And in that one, I would want it to count towards the same pool of like This is all the stuff you've contributed and you get rewards based on that But what are those rewards? I don't really know. Um, you know, something like, um mentorship might be a way but at the same time letting people buy time with me is also problematic for other reasons But then I start thinking well, what even are reasonable rewards. I'm not entirely sure I don't it's not like I have art that I can give people for example, right? So if you've I don't know if you've sent me like a thousand dollars What would you get back from that in some sense, right? The the weird thing about what I do is that I The the thing that I would get money for is the stuff that I've already done or I'm already doing anyway Things like the, you know, cross for cross for rust or the other rust videos or the russetian station podcast Or the open source work that I do or the missing semester course Um, or you know follow-up work on rust for rustations Like all of that stuff I'm going to be doing regardless But the there's no individual reward and there's not really a The more you give the more reward you get So there's just something weird about the model here that I don't quite know how to how to Match with the real world. Um, so I I don't know what that is all going to look like Which then gets back to the question of do you plan on a community discord? Which is Yes, but I don't know how I want to do that yet either and I think it's going to tie into this somehow so for example, I might start one where there are like Let's say four tiers of channels where one of them I have notifications turned on one of them. I promise I will check once a day One of my promise I will check once a week and one of them I'll just do whenever I spare time and the more you've contributed the sort of higher tier channel you get to be part of Maybe I don't know. Um, but I also don't want to start something prematurely and then Change it significantly later. That also feels weird Um, so I don't But but but I have it is something I think would be fun It would be fun to have a discord specifically for the sort of community around my streams or more my content Um, I just don't know the right way to do it Um So that's my take there Uh Next question What are we at time wise eight? Okay, we have some time remember to go down to the questions near the bottom and uh vote on some of those because we have a bunch of questions down there that have been asked Recently that have very few uh votes and the ones at the top are the ones that have been there the longest So go down every now and again and vote for things you think are interesting Or I guess really go to the middle because those are the ones that people have other people have voted on But not enough to get it to the top Um What are your thoughts on compiler backends? Uh crane lift and zig's new ambitious plans to remove dependency on lvm and clang um I think it is good that people work on additional compiler backends Stuff like crane lift very cool Some of the zig proposal to get rid of lvm Seems ambitious but cool at the same time Um I don't know how much it matters Uh, I also worry that it is um It's essentially diluting impact right it means that Out of the relatively small set of people who are good at working out like compiler internals, uh, and Low level like generating machine code and stuff That talent gets diluted across many different projects rather than being focused on a few One of the things that lvm and clang really succeeded at was bringing together The community of the people who care about and know those things and and By splitting that again, I feel like we're Going backwards. Um, I I acknowledge that there are some there's some value in having more custom more I don't want to say more or less over engineered that that's maybe the take from the people building these but like I think it's useful to have frameworks here. Um, rather than just custom building everything for the language you build on top. Um But at the same time, I think it's valuable for compilers to support multiple compiler backends Um, I don't know that building a new one for yourself is the way to do that I would really like to see a Like a rust version of lvm and clang But that's also a much longer pull. Uh, and that I think is valuable Only if you don't end up being like the n plus one project But instead if you actually manage to capture A bunch of the talent that's currently in lvm and clang and bring them over so that you again So that you don't dilute the um the value here um Okay, uh, what would be the salary of a principal engineer in the eu I've already talked about this Uh, have you finally moved to wayland? If so, what has been your experience? Yes. So with this new computer I for the first time installed wayland, uh, and set everything up on wayland and it was a Um It was a mixed experience. So some things worked just straight out of the box, which was really nice um, some things were just Weird concept wise like the fact that the Window manager and the compositor and the desktop environment are all the same thing or was Confusing and I had to rack my brain a little but at the same time it's pretty convenient. Um Some things around how inputs are handled are pretty weird Things like global shortcuts and obs not working has been annoying. Um Uh and and other kinds of integrations like I have a status bar at the bottom of the screen um, and I wanted to show all of my workspaces, but Hyper lens definition of what workspaces are It doesn't expose those in a way that the bar can pick up and there's doesn't seem to be really a standard For how to do that through wayland and so as a result I just have to build a custom integration because no one's really built it before and that's Annoying there's a bunch of things like I found a bunch of different bugs and different pieces of software that don't work quite right with wayland like button handling on my drawing tablet Um num had keys for shortcuts um What else did I run into? um At this bar that doesn't work quite right my drawing program Will sometimes randomly hang its color picker, but only under wayland stdm. Just gives me a black screen on wayland. So there's like It's clearly the less battle tested I don't know what you call wayland or x11 windows server It's clearly less battle tested than x11 is But at the same time, you know, I do think it's the future I don't think we're gonna stick on on x11 forever either Which is one of the reasons I moved was like the only way to make the move Nicer is to do the move and then figure out what doesn't work and file issues So I've filed a bunch of bugs over the past uh week or so with things that doesn't work Um, yeah, so I'm doing I mean, I mean this is the obs now is doing screen sharing is doing The webcam and everything just do wayland and it's working just fine Um Uh What are the plans on more crust of rust? There will be more crust of rust. Um I don't have any concrete plans that the challenge with crust of rust is finding the right topic So crust of rust is not just like Let's talk about random rust things rather crust of rust is trying to find what are Core pieces of rust knowledge that need to be disseminated more broadly or discussed more broadly and I don't have A very long list of those topics anymore. I feel like I've done most of the crust of rust that I wanted to do There are probably more. I just i'm still working my way through what those should be Uh, which is all to say like there will be more, but I don't know when and I don't know what the what they'll be on Uh, what rust rfcs. Are you excited about this one? I'm actually gonna skip and the reason is because Given all the moving and everything. I'm not really in the loop on what rfcs are currently ongoing that there are none that I'm Monitoring particularly closely. Uh, and I feel like I'm out of the loop on which ones have actually made meaningful progress Uh, so I'm gonna skip that one What is your setup now? I'll mention it briefly. I think I'll do a um I'll probably do a new like setup video, which I know I've been promising now for like four years But now that I've moved to Wayland, I feel like I really should because my setup is pretty different but very briefly my setup now is Hyperland as the desktop environment um, and window manager Yambar as my bar um Rofi as my launcher Firefox as my browser Still alacrity as my terminal running tmux with neo vm Uh, neo mutt as my mail client Uh Still using rust analyzer with neo vm Still using the grovbox dark hard theme and the notosans font um I think those are like the main bits of setup Yeah, I think those are the main components of setup at least Uh, oh and one password for password management, which I think is also a change from last time Yeah And that's it Uh, okay How can someone with three plus years of experience but no formal computer science background effectively begin learning or working in distributed systems? um Just apply to jobs the formal computer science background shouldn't matter if you have three years experience um Again, I think these companies often Nowadays at least the requirements that are listed on job postings is usually bachelor's degree or equivalent Uh, where the equivalent can be years of experience. So usually like When you see requirements listings for jobs, think of them as guidelines not as requirements Um, unless it's something very specific, but but very often it's it's much looser than that um Reach out in like worst case. You don't hear back um But but like if you have three plus years of experience then that Is a good way for you to get into something Um, and you should be a good candidate for for jobs I don't think you need the formal computer science background Do you think rust jobs will increase in the coming years? Yes by a lot. I think rust is Significantly on the rise and has been for a while and will continue to be for a while Um, and I think that it is a booming job market Rust is that is so very much so Whose voice is it in the intro to crust of rust when it says um, num, num, that's mine Uh, I leaned really really close to the microphone. I went um, num, num So that's me Uh I'm just going through these somewhat more quickly now to try to make sure I get through more of them Uh besides rust what other programming languages will you use in this new job? Well, so I don't fully know yet. Um, because I haven't looked too closely at their tech stack But I know, you know, this rust there's obviously python. There's obviously bash because there always is in any real system They use nix for some things and I think they want to use it for more So I'm probably gonna have to learn nix language. I'm a little sad about but it's okay I don't know that there really be any others In anything that's large like this and especially when you have to interface with other systems Um, there is basically always some c or c++ somewhere in there. Um, and so if there is I'm going to end up interfacing with that But not that I know of So a relatively small set And maybe some open cl or CUDA if it really gets down to it, but I'm hoping to avoid that as much as I can Uh, what are the requirements to be a principal engineer? Let me send you Uh Here there's a really good article by um, ilia gregoric called the principal engineer's handbook um and this one is I think a is so Ilia is a principal engineer and the technical advisor to the ceo at Shopify And it's a really good just like overview of what does it mean to be a principal? And you know, it varies by different companies. It varies by the size of the company. Um, so there's no One size fits all definition but The way that I think about a principal engineer is as someone who is relatively self-directed um is sort of a Searcher that that looks for where problems should be solved where they're where the highest um impact highest priority issues and ghost seeks them out and either works on them themselves or escalates them Or interfaces with the team to bring things forward. There's sort of a um A person who Surveys reports and attacks as necessary So you do both sort of broad and deep um, it also does You know in terms of requirements it requires that you're decently good. I think at working with people not in a people management capacity, but just in the sense of um You're going to have to want to are you going to want to and have to talk to a bunch of different teams a bunch of different engineers about What works for them? What doesn't? What is what are barriers for them? What's friction where have they seen big wins that might be able to take elsewhere? you have to like dig and learn and understand and and um And figure out what matters and what doesn't what identify what thing generalizes, which ones don't um There's a decent amount of requirement for just expertise and like hard one wisdom Of you know that these kinds of things work. You know that these kinds of things don't you know The rough shape of a bunch of different problems and how to generally attack them um, I think there's a lot of judgment too like and I mean judgment in the sense of like You're faced with trade-offs and you have to make trade-offs and you have to make them in an intelligent way based on data based on past experience based on sometimes guesstimation But but you get put into positions of Difficult non-obvious trade-offs more often and you need to be comfortable and willing to navigate that space um and Yeah, apart from that, you know, I think read ilia's handbook. It's it's a really good intro to Roughly, what do you do but also what's required? um Do you think rust is going to be A prominent language used to build ai systems in the coming future taking over python. We've talked about this earlier um Popular autograd libraries like pytorch and tensorflow use c++ for hardware kernels What do you think needs to be done to get rust to be the default choice? Do you think this is a good domain for rust? I think this would be a great domain for rust To sort of replace the c++ parts of pytorch and tensorflow um I think part of the challenge there is the tide integration with Like open cl and kuda which are both pretty hefty c++ things from from the vendors and so It's going to be hard to bring rust in there without some buy-in from the vendors of like kuda in particular Like nvidia needs to get on board with rust for I think for rust to really succeed here um And I I think it's not unrealistic that they will um But I I do think rust would be a great thing to get in here And I I want c++ to Disappear from that space to the extent that we can but it's going to be a a long slog I think Um Have you considered making a video on ghost cell q cell or similar libraries? I've thought about it They're a little bit niche still like they use some pretty cool tricks but in practice their applicability is a Little less than you might think um There are some places they work really well, but they're not broadly applicable even though I really wish they were um Some of them are but but in practice like you just don't need them all that often um That's not to say that it's not worth Going through them like I think it could be an interesting like deep dive on unsafe for example Like an unsafe chronicles type video Um, but it's not something I've currently planned. It's like somewhere in the far back of my head Um I'm looking for low level development resources for quant development and rust can you please suggest some resources? I don't know what you mean by quant development But if you mean quantum then no, uh, if you mean something else then also no, uh Not something that I do or have insight into Um, you've been an engineer for a long time for someone starting now. What are the pitfalls to avoid in your opinion? um Oh, this this stuff could fill a book. Um I think one thing that is valuable as an engineer is to have some degree of humility and that doesn't mean Not standing up for yourself. It doesn't mean not having confidence in your own solutions but it means a willingness to listen to others and to Figure out whether they might actually be right and you might actually be wrong I see it far too often re-engineers are so Stuck in the mindset that their solution is the right one or that their way of looking at solutions is the right one that they miss Early warning signs. They miss obvious errors. They miss huge opportunities for gains just by being stuck in their own way of thinking um and I think having that humility and Nurturing that over time is pretty valuable, especially as you become more senior because as you become more senior you you In some sense, rightfully feel as though you know better But you got to you got to fight that you got to put that intention with the fact that other people still have valuable ideas like Other people who are coming up are also going to become senior and have good ideas that might be different from yours So that I think is one of them um The other thing that I think is really valuable is to start to Track your own work and I don't mean that in terms of tracking time I mean that in terms of tracking What did you do? right like your accomplishment not not just at one company but across your career as well um, there is a Uh, where is that? uh, julia evans Wrote this blog post called get your work recognized write a brag document And it's a really good blog post that basically has the the top tagline You don't remember everything you did and this comes up again and again It comes up whether you are like going up for performance review. You're going up for promotion You're looking for a new job. You're obtaining your resume. You're applying for Like further education. It can be all sorts of things But it's just really useful to have a list of like cool things you did because you will not remember them all It's good for yourself in terms of others, but it's also good for yourself It lets you celebrate and recognize and remember your victories Um, and you know, hopefully gives you a sort of track record of like, you know, I'm actually pretty good It's a great way to battle imposter syndrome. Like start keeping that kind of log pretty early It's something I wish I had started earlier because for me There's a bunch of stuff. I'm sure I did and was proud of a long time ago, but now I don't remember many of them um And and I think maybe the third thing I would say is Try to not get stuck. Um It's very easy to become Comfortable with or good at a subset of your field and then just sort of do that for the rest of your life And if you're happy with that, that's fine. Um, but it's um It can be a Dangerous kind of comfort because over time, you know technology develops People move to different tech stacks new things become the new hotness um People move new companies start to spring up in new domains old companies fold The way that we build software changes the world changes. Um, and If there's just one thing that you know how to do You're going to trap yourself into the only other places you can go or places that are sort of adjacent to that um And so if you can expand your horizons a little bit like try to make sure you learn new things and and Then then that's going to benefit you massively because it gives you a bigger surface area to contact with other opportunities and That doesn't just have to be in terms of pure technology. It doesn't just have to be like learn a new language It can be learn a new field like learn, um Biology chemistry computational biology is the thing right it could be physics. It could be math. It could be formal methods um, it could be like physical like engineering, um, it could be 3d modeling it could be digital art. It could be just art in general. Um, it could be, um Journalism it could be psychology It could be a ui design ux design qa testing All of these are hugely valuable skills and again, it doesn't have to be related to this at all It could be law for that matter and you will find ways for it to then merge with the things You know from before and now you're a more unique candidate with a more unique set of skills And it also unlocks other things that you can work on. Um And that that ability to explore new spaces and like I mean again This is one of the things I think the phd gave me is that ability to explore and build up a more robust and broad Foundation of skills and interests and I think that benefited me a long time and still does um Do you think rust will beat c++ in the robotics and system engineering space? Yes. I think it already is um, you know, it it it will take some time but Just because there's a lot of software in the world. Um, but but I think this is already happening I again, I don't think there are any places where you should use c++ if you have a choice Now there are a bunch of places where you don't really have a choice. Um, where for example, um Where There are vendor provided libraries that only have bindings in c++ and so using a different language is really painful Granted those are more painful to work in but and then robotics for example Like the framework for the os for the device you're programming for might just be a c++ thing That you need to integrate with and in that case Your choice your choices are more limited But on the whole I think you choose rust over c++ because there will be a better choice now and in the future um Yeah, finance is another thing to be really it's worth learning um Do you think rust c will ever move to an id oriented architecture sharing its code base with rust analyzer? I don't know I feel like there's been a lot of discussion in both directions here Um, and I feel like there was a push towards this Then it sort of failed or didn't happen And then I feel like there was some conclusion that maybe we should just Not try to unify these and instead it's better to just have the have rust analyzer be its own separate implementation of the language I don't know where it's landed and I don't know what the right answer is either like this I think requires someone who's like a Compiler expert and also an ide expert which really just means matclad needs to do it But also I think matclad is now doing zig stuff Um, I don't know. I don't have a great answer to this speaking of skills that engineers Should make sure to learn that are important as you become more senior like pitfalls is Learn to tell people when you don't know Because it's so easy as engineers to just try to make up an answer on the spot And you need to just acknowledge when you don't have a good answer Because otherwise you're going to give an answer and other people are going to take that as not necessarily as gospel But as this person knows what they're talking about because they gave us an answer Try to avoid that Um, did you experience imposter syndrome when working for AWS or your recent interviewing? Um No, but that's because I'm weird Um I've never really had a confidence Problem quite to the contrary um to the point where I very much had to learn humility growing up and eventually I did um And now I'm the best in the world at humility. No, um, so for me I've I've not really been affected by imposter syndrome all that much um But I know that that is the rare case um What is your opinion of rust from embedded platforms like arduino and raspberry? I think that's I think that's a great idea All embedded platforms should all be rust Um, how did you meet girls when you were a phd student? Um I was on all of the apps and sites is really the answer And You know the same way as other people I guess but but interestingly enough like, you know, I was talking about how um, when you do a phd you have a lot more just time and room to explore and Have spare time and in my case one of the ways to do that was to try all of the different Dating apps and platforms for this. Um, so I think I had a profile on everything and was I don't want to say active that that sounds Uh too severe But but rather like I was like testing them like I was Is trying to figure out which one worked best for me and what works best for what? Um, and ultimately I didn't I met my partner on one of them. I became my partner on um Hinge in the end, uh, but I think that was somewhat arbitrary. Um I think In some sense my favorite was actually okay cupid because I like that you Give a lot more detail. You have more to go on with people. Um They have all these questions you can ask and then see Your match with other people based on not just what you answered But what you have said you would like the other person to have answered and how important it is to you So I like that set of better Um But not where I met my current partner um hinge were decently well for me mostly because I or when I say worked well, I mean I liked it more than tinder and bumble for example because um, it was set up so that You have to start the conversation with something So like when you first contact someone you have to like either pick a picture or a particular part of their profile and send them a message about that And that I think was a A forcing function that that was pretty nice um So Yeah, I think that's I mean follow-up questions are welcome, but I think that's the rough answer. Um, and then you know, I Ran some board game nights and stuff and that's a decent place to meet people. Um But ultimately, you know, it's hard to find someone that you want to be with longer term It it's not It's not just mere exposure, right? There are a lot of things that need to go right And so I felt like okay, I need to go out and search and That so then searching is a bunch of work um What whale and compositor are using hyperland I don't know if that's how you pronounce it hyperland It's h y p r l a n d There's no e and I it's very frustrating because of the command for configuring hyperland is uh H y p r c t l like hyper control but Every time I type it I type it with an e like h i p e r c t l and every time is wrong and I have to type it again That's really annoying He propellant Maybe it has to be said very fast Uh, what would you change about your job? Well, I don't know. I haven't started it yet Um, it was going full time with commercializing noria something that you were considering. Um Well, so I I co-founded the company to start commercializing noria ready set and The main reason why I didn't start working for that company straight away was because I um I was so burnt out with database work Where I'd worked on noria for like six years straight day in day out weekends Like that was just burnt into my brain and so I was I just didn't have the The capacity to do any more work on it. Um, I still think it's important and valuable, but it's just Certainly at the time when I graduated I was just Even just the thought of going back into the codebase made me be unproductive no matter what I was working on Um And so so I recognized that the right place for me to go would not be to work on commercializing noria that it just wouldn't work But I think it was important to do that and hence I co-funded but did not work for It's gotten a little bit better, but even now like I still feel this Tingling of like it's too it's too soon like databases and especially the noria codebase is still like So burned into my memory that I need distance from it Um Have you tried using emacs? I tried emacs for a little bit and I didn't really like it Um, one of the things I didn't like was that it's not really a terminal editor It wants its own gooey and I don't want to give my editor its own gooey Um, and the other reason was because the key bindings were weird Um, and the third reason was it felt like It was more an operating system than an editor and that's not what I wanted Um, those aren't great reasons. Um, I don't have like a Burning passion that people shouldn't use emacs. I just that just wasn't for me at the time at least um What is your advice for people struggling To find a job in the tech industry nowadays not necessarily as a rustup It makes me think that the industry is still saturated and I should focus on personal projects instead Would like to know your thoughts Cheers man signed just a dude Nice sign. Um I don't know that the industry is saturated. I think the industry is maybe saturated for More junior engineers, perhaps Um, so it would be useful to know more about your Um, your background here to really answer this well But but I do think that there are a lot of tech jobs out there Uh, especially in rust for junior and senior people. Um, one of the things though that I found that is unfortunate is a lot of Especially startups Tend to want to try to hire senior people and I think that's a mistake I think if you build your company with only senior people until you get to a size where you're like, okay Now we can now it's okay. We can start hiring junior people. You build a senior Senior engineer culture. You don't build the mechanisms the devices the tools the processes for Making junior engineers succeed at your company and turn into senior engineers I think if you build the company with Junior engineers present from the start it forces you To build the company in a sustainable way where you have a pipeline for engineers to go up through the company Um, and so I I wish there were more tech startups that hired junior people Uh, which brings me to the the obvious follow-up question, which is I don't know what helsing is doing here Um, I believe that helsing is pretty good about hiring junior people as well I don't know whether they're they currently have any openings. It's worth looking um And if they don't have any openings for junior people, then I want to have a word with them and see But but this is certainly an area where I think it is very valuable for companies to hire junior people But I don't at the same time. I don't know what level you're at and I don't know Which languages you're looking at or what country or region you're looking at and whether you're open to remote Whether you're only looking for in person Um, it it's difficult. Um, also also Companies tend to mostly make exceptions in terms of things like remote for more senior people So if you're junior you have less availability for companies because you have to look at companies that are either fully remote where everyone can be remote or That are on location where you are and that limits your search as well um But if you're struggling to find something I do think that focusing on personal projects is pretty good Although keep in mind that that mostly helps you for self development right so Like it'll make you better as an engineer. It's not clear how much it'll help you on your resume like open source work is Tricky to use in a resume setting because if you're on the Receiving side of a resume that mostly lists open source stuff Evaluating how well you ran open source projects and what kind of experience you've gotten from it is tricky at best so you will generally probably Not get that much benefits from it resume wise um Does norway have a lot to learn in terms of tech industry should the Norwegians take more risks and be more innovative? um I don't know that it's about risk or innovation um I think it's partially the norways just really small. They're just aren't that many Norwegians And so there's fewer of us to be innovative. It also means that it's harder to start a company because you're initial audience size your initial customer base is smaller um The taxation rules in norway are also so that such that it's Pretty costly and annoying to start a small company um Which means many people don't or they start at the mouseware I know that the rules have gotten a little bit better Recently like in the past few years. So maybe that starts to change things But but I don't think norway is non innovative I think it's a combination of those factors that just make it that we don't have like huge booms coming out of here I also think there's a pay problem and this applies more broadly in europe but but certainly in norway that I I had a couple of conversations with Norwegian tech companies And they weren't really competitive in terms of salaries, which means you're not pulling talent here Which stifles your innovation and also stifles the possibility for future inventions here Um But but I don't think it's about risk-taking or or being innovative in the first place Um Do you know who magnus carlson is if so how famous is he in norway? Yes, I know who he is and he's quite famous in norway. Yeah We're we're very proud of anyone Anyone who's from norway who does it well in the world because there aren't very many of them There aren't very many of us And so we get irrationally excited when we see like a new region flag somewhere else in the world Very cool. If there's a new region flag like on tv or Like if I watch an american show and in the show, there's a new region flag I get a little excited or there's a new region in there excited because we're just we're just tiny. There aren't that many Um What do you think of free bsd and all the other bsds? It's also it'll also be great to hear what you think about next to us Um, I like bsd. I ran open open bsd for a little while the biggest challenge. I think I had was that It was just too inconvenient like I wanted too many things that were linux only and so I just gave up Um, maybe things are better now to the point where I should consider trying it again. Um, but I don't know. I know this this also gets to the I don't really feel the The carrot or the stick, right? There's nothing that's really pulling me to bsd And there's nothing that's really pushing me away from linux. And so I'm pretty happy where I am um next to us I like the idea There too, I don't really know what the incentive is to move I've heard that it is Easier to get things to just work because they're installed as as bundles rather than as you know You build your application to dump it on your file system and that appeals to me to some extent um But I've also heard that the nyx language is a huge pain and I quite like writing arch package builds, which are just bash um, so I I like the idea of nyx os. Maybe I'll start tinkering around with it Given that helsing does some nyx stuff as well. It's not unreasonable that I might end up, um Uh Dabbling in it and then it infects my entire life, which it seems how is how nyx gets into people's lives It starts as a little bit and then it just takes over Um, but that's the extent to which I have uh useful input there Um What are your thoughts about working in the security field security is really fun, um, I mean it's terrifying But I I really like security. I was the the teaching assistant for the computer security class at MIT for many years um, and it's I think it's fascinating to think about threat models to think about security mechanisms and they're the trade-offs involved in the space the trade-off between user convenience and um and security Um, I I think it's a hugely interesting field. I don't know what it's like to work in the field now At amazon and probably at helsing too a decent amount of the work that I do is security Related or has components that are that are about security But working as like a security engineer, for example is not a position. I've had but I've interacted a lot with them um, and I it sounds Really fun But also quite stressful because you are on the hook for getting things right And when things go wrong, you're the first person paged And I'm okay with that not being me But it seems really interesting to me Have you tried leptos for front end development? And what are your thoughts about front end development in rust in general? I have not tried leptos and my thoughts about front end development in rust is that um It's weird to me to do front end development in rust because rust is not a front end language I think rust makes a lot of sense for Web assembly like when you want to write specific components that require the sort of high performance or or correctness Stuff or low level control that you get in rust But building your entire front end in rust feels a little weird It does depend on what you mean by front end because if what you mean is sort of The the thing that generates the html and css Then like that's not really front end development to me There there is some of that maybe and that's where I think something like leptos Falls in but when it comes to like building something like single page apps or reactive applications and stuff I just wouldn't do that in rust. It feels weird to to force that to be in the sort of Back end that then gets converted. I think that's a recipe for pain I do think that rust has a place Relatively close to the front end. It still makes sense In in terms of things like generating html generating, you know, the htb Sort of service kind of thing that there makes a lot of sense to me But in the actual development of the ui probably not or at least it doesn't make sense as much sense to me Okay How will tech look like in a decade I Don't know probably about the same as now I just like don't believe in significant radical change over the course of only 10 years It happens but like More ai If I had to make a like a radical guess, let's turn this question interesting like What is the most radical guess I can make as to how development will be different in 10 years from now? um I think that the current trend to work remotely is Not slowing down it is accelerating and I think that uh We're going to see a Not just remote first but a remote only approach to work Where fewer and fewer people actually go into offices or want to go into offices And I think it's going to Completely change how companies think about their about engineering culture about engineering how we do remote engineering because A lot of the practices that especially the big tech companies use are very much rooted in how we ran development During again person type things and I think it's just going to be radically different like things like um real-time collaborative Coding I think we're going to see more of You know there are some editors are starting to get this now where you can have Multiple cursor in a code editor and I think that's the kind of direction that we're going in Where like you know in in the olden days you would sit down at a computer together and like pair program And I think that's all going to be remote. I think we're we're leaving the world of in-office software engineering behind And there are a bunch of things that are hard about that and that that we don't know how to solve But a decade from now, I I think we're going to see that trend accelerate Uh What back-end framework do you recommend axiom, actic salvo or any other I like axiom um For a bunch of reasons, but none that are particularly well informed except that I've used it. I like it I know the developer pretty well and Or I've interacted with them sort of online a bunch and they built good stuff um Integrates pretty closely with tokyo, which I think is a good idea. I like axiom Uh Do you think rust is the best language to different implementing cryptography right now as opposed to cc plus plus they can go? Yes um Go is go is very good at cryptography right now But it's mostly because they have really good cryptographers working full-time on that problem I don't think it's because the language is inherently better suited for it. Um I still struggle to see what zig's unique selling point is or why that's better than rust But you know, I could be convinced wrong there But I would rather them be built in rust like anything where you don't Need something that goes beyond the boundaries that rust puts on you. I think is good to do in rust um All right, I'm gonna speed run through a little bit here and aggressively skip some questions Um, because I want to end it probably around nine. So in like six minutes. So that's three hour mark I think it's about good. I'm gonna I'm gonna run through um skip that one skip that one um I've been catching up on streams from a few years ago Can you explain why you didn't go back to the lock free to weight free simulation after implementing hazard pointers? I just haven't had the time. I still want to get back to that stream I think it's really cool and interesting and the whole point was that once we get hazard pointers We can actually make progress. So I I genuinely want to go do that more. Um, it's it's on my list um What made you leave the us? Oh, there's a bunch of things. Um, I left the us because Um, I think the way that healthcare is handled there is absolutely insane um, I don't like the car-centric cities and the car-centric culture of having to drive everywhere um, I think the intrusions on civil rights in the us including things like the right to abortion being taken away is is wild um I um I also think that the fact that Guns are a problem in the society at the level that it is is not something I want a part of um And the fact that you have basically no employee rights in the us You know you not only it's like healthcare tied to your employer But also like you can be fired on the day. There's no required notice period You have very few rights as an employee union culture is sort of dying out um And taking together like I just don't see why you would stay in the us right now. I just don't think it is valuable um Why do you choose to work in AI and not some high-frequency trading quant firms? Um I his quant again Um I don't know I I I don't Work at Helsing because of the AI stuff the AI stuff. I don't really care about it's an enabling technology That's used in part of the stack I did it because a I think what they do is important and b because they have some really unique problems in the distributed system space High frequency trading not so much um like Sure, you need to be low latency in high throughput and whatnot, but like They're not uniquely different interesting problems to me um Do normal developers need to care about the fuss around the rust or again the trademark issues? Does that affect anything if we're just developers? Mostly no. I also talked about this earlier um Already answered that What's your opinion on sabbaticals? I think everyone should take more breaks and I think that should be a standard part of our working culture I think we should work fewer hours. I think we should work fewer days. I think we should take more vacations longer vacations Um, so sabbaticals very much on board. I think it's a great idea um Do you have any thoughts on aida compared to rust? I I asked I took this question just because of the use of the person who posted it um But no not really. Um, I haven't used aida at I want to say I used it once for some random task, but I don't really know aida Um, so I can't really speak to the the comparison, but I like the username you put um How do you feel about the fact that nearly everything is close on sundays? So in norway, a lot of things are close on sundays including most grocery stores. Um, it's a little annoying, but at the same time You know, I like the fact that people use the day to do other things instead And more people have the day off. It's kind of nice Um As you get older, do you think your ability to solve problems or learn increases or diminishes? I think my ability to solve problems is increasing. I think my ability to learn It's probably about the same It's changed. I think my ability to just like Do retention? Um of knowledge has probably gone down But at the same time I have more hooks to hang that knowledge on So It's easier for me to pick up new things because they relate to things. I already know so a little bit of both Um Is there a role for rust in the field of ai or is that ship sailed? Very much a role for rust in that field Um Now that you're no longer at aws or amazon, what's your opinion of the company? It's management techniques in particular like stack ranking employees Um, I think stack ranking makes absolutely no sense. Um, this is the same thing that certain universities do and with, um Like grading on a curve. It makes absolutely no sense. It's it's truly ridiculous Um, I understand why they do it because it's easier and more convenient, but it's a terrible practice. Um In terms of the company overall, I I think amazon Get some things really right like I actually think the leadership principles that amazon are really good um The two new ones are not so good, but the other ones are quite good. They're sensible And if you apply them well, they're a useful coherent Vocabulary to use for internal conversations and discussions And for guiding those and for guiding documents. I think the document culture is good too Um The problem of the leadership principles is that they're not universally followed They don't seem to really apply to leadership and those are senior leadership in the same way it applies to the rest of the company So they end up being sort of used to downwards, which is really annoying Uh and unfortunate But but I think you know that I think they have some pieces right I think where they're really going in the wrong direction now is on remote, of course Like I think I think they're shooting themselves in the foot. I don't think they have much of a choice For what it's worth, but but I think they are um and you know, I think amazon feels a little bit now like a two-tier company like there's sort of the There's a senior leadership And then there's the rest of the company and if I feel like there's a disconnect There's probably a communication disconnect and I think it might be partially because of coven Where they relied very heavily on in-person communication to stay in sync with the rest of the company and that's sort of gone away uh, and they haven't really Remedyed that situation except now trying to push everyone back into um in person um but I'm not Hyper critical at amazon even after leaving. I I do think that the I don't like the way that they treat employees as sort of fungible and to some extent disposable um that that Really rubs me the wrong way and the way that they lay people off and such um So mixed feelings there um, but I I value the time that I had there very much um Let's see How's your experience been living on a visa and how'd you get a work permit? um, the work permit I got Well, so what happened was I was I did my phd in the us and after your phd you can do what's known as um practical training So you get a practical training visa and that lets you work for One to three years um on that Student visa which makes it easier for companies to hire you and see how you do because they don't have to sponsor the visa And then after I joined amazon, they then immediately started applying for an h1b for me. Um and I got that like a year later or something and um That process wasn't too bad Um for smaller companies, that's more annoying. They're not if they're not used to sponsoring visas And that's the whole hassle I did not like being on a visa. I I think it sucks. Um It's one of the reasons why I moved back to norway is so that I don't have to be on a visa And certainly not an employer sponsored one It puts you at a disadvantage to the company for sure But in addition it puts a bunch of restrictions what you can do with your spare time Like for me, I really wanted to like open a patreon. For example, not allowed um I wanted to do some rust consulting not allowed. Um It is just there's a bunch of things that you're just not allowed to do when you're on One of those visas and that was Very very frustrating um Let's see I'm scrolling through here. Are you still learning the piano? How's it going? Um No, well, I I was a little bit but my piano stayed in la and I rather I sold it in la And I haven't bought one here yet and I don't really know where it would go The apartment is a little too small But I do have this is a guitar that I bought many years ago in london and I found it when I came home Um, just like a few weeks ago So maybe I'll now try to learn the guitar again. I don't know um Does housing offer visa sponsorship? Yes I mean they do um That there are some restrictions on uh Because they work with like the military for different governments and so that there are all these Requirements coming from there where There are certain passports And dual citizenship statuses That they cannot hire like they're just not allowed because of the clients that they have So there are a couple of restrictions in that way But as long as those don't apply to you, they do offer visa sponsorship and I think they offer relocation to So you don't have to currently be in europe, but you will have to move to europe if you're not in europe I'm just browsing through here Are you interested in joining rust meetups in oslo? I already answered that one But yes, I am I'm just waiting for the next one to be scheduled um Do you speak Norwegian and if not how is living there with english only? I would like to emigrate to another country, but language is the biggest problem for me. Um, I do speak norwegian. I'm from norway myself Um, but my partner does not uh, and I think she's found it surprisingly easy like Everyone speaks english here and that there's a surprising number of people now just walking around the streets for the past You know a month or so that are also english speakers like just speaking english on the street um The main problem places where you might have problems is like if you travel far outside the cities Or if you talk to much older people, they might be less comfortable speaking english But in general people are pretty excited to speak english here, too. They speak it pretty well Um, so you you won't have too many problems with english only here um I like to play videos on 1.5 to 2x speed. I think it makes me speak faster I've noticed that you speak very clearly and steadily. Is it something that's natural to you or do you train to do it? um I that this is how I speak and and like I don't know whether it's morphed over time as I do more videos, but like I think I Try to be very intentional with my words when I speak normally And so I do that when I speak to the camera as well. Um, and that Means that I end up speaking clearly and steadily Sometimes I speak quite quickly But I think even then I try to make sure that I use all the right words as I go And and so it still ends up being very The words end up being very segregated from each other Which I think is part of what gives that that impression. Hopefully that makes it easier to understand me at higher speeds, too Um Let's see. I'm getting down to like the four vote questions Uh, where can people who do not use twitter follow your stream announcement nitter.net? Unfortunately, it doesn't work anymore Um, I'm also on mastodon uh, fosterdon.org slash at john who um, there's also a Um a calendar with all of my streams so you can just You know, um Follow that calendar That link is if you go to my fosterdom, for example, you go back a little bit You'll see the link to where I announced that I'm going to start. Um keeping a schedule Um, and apart from that, I mean this is one of the reasons why I want to set up something like a community discord or a Uh, or a patreon or something is so that it's easier for people to just get announcements from me Um, I'm also now starting to post on linkedin. So that's the other place where you can follow me Don't connect to me on linkedin. Just follow the connections are different and annoying Um, so that's the other thing to use Um, let's see Can we see photos of your sourdough? No, I threw it out. It was very very sad Um, but I will make sure to post some when I when I uh managed to get it to work again Me dumb how me get smart It's a good question All right, this list of questions is now Very long One of them just says, how are you doing? Which is a good question. I'm doing well. Thank you Uh, I'm trying to look for the ones that have been asked Recently that are the ones from chat Um Just to serve the people in chat here better Um What do you think about restation station? I mean I help run the podcast. So I think it's a pretty good thing um recently I haven't done too many episodes on it because of the move and everything but I'm hoping to get get back to that We're working on windows os be a deal breaker. Yes. I have no interest in working on windows and there's a difference between If I'm building software that other people run on windows and whether I have to develop on windows That those are a little bit different. But yes, I do not want to be forced to work on windows How would you compare working in the us versus norway? Would you recommend it for a new region? I think it was really Valuable for me to work in the us for a few years. Like it's not as though I would undo my time there it is more that It's not where I would want to be long term um I don't know that I would I so I would I would very much recommend working in the us for a little bit if you can at the same time like I don't really like the us and you would have to put up with working on the us for a bit, which Is it worth it hard to say? I would probably work for a european company instead whether remote or in that country Um Which neo vim front end or ui to use with wayland and if it's not G n vim, have you tried it? I don't use a front end gua. I use the terminal ui So I launch the lacrity. I have tmux running it and then I run N n vim directly in my terminal It's the only ui I've ever wanted to use Um, okay, I'll do Do Two more questions So let's I'll I'll start from the top and if you vote something else up then uh, then we'll take it from there I have not tried obsidian md. So I'll just get rid of that question. So it's not a um A candidate here. What started your interest in rust? Um So my interest in rust started um Really during my phd. So from my first the first year of my phd. I used to go Um, and it was fine. I didn't mind it. I like the language well enough um But then I decided to start building this database like thing and you know, I knew it was going to be High performance low latency. I knew it was going to need to do a bunch of like memory tricks. Um, I knew that it was going to be Fairly like low level code, but also I really felt as though I missed types in go Um, or I wanted to do more of them And while all of that was sort of circling in my head and I was going to start writing the first prototype I was like, all right, I'll use go for this as well. Um, and I wrote the first prototype and It was fine and then I saw the announcement of rust 1.0 and I went like, hmm this thing proposes a language That or gives a language that does a lot of or has a lot of the properties that I'm looking for in the language for this particular problem I'm doing a phd. I can just choose whatever tech stack I want. No one else is going to tell me not to um And so I just went with sure let's try rust then I guess Um, and I just never really looked back. I started using it and just really enjoyed the experience I started bringing other people on board and they enjoyed the experience after the initial learning curve It worked well with multiple people working on the project. It worked well for several years Without, you know, significant regressions and without any huge frustrations with the language And so it it was really one of those It it fit the problem that I was looking at at the time And I had the freedom to allow me to test it out and then that test was very successful and then the tests since have all been successful um So there's nothing really more Fancy to it than that. It seemed like the right tool for the job and turned out to be um All right last question What dynamic disc what can dynamic dispatch do that static and enum dispatch can't? Um Dynamic dispatch do that static and ema nenum dispatch can't um It depends what you mean by can do so The first thing that comes to mind is that dynamic dispatch Um Let's you type a race So you don't have to pass around a concrete type um with a with static The you know, if it's monomorphized the type is relevant for enum dispatch You need to have an enum variant for the thing you want to call um with dynamic dispatch you don't have to name a type um You name the trait you don't name the type and this matters in the sense that With static dispatch Anyone who calls you or your caller can choose any type that implements the traits With enum, uh, which is nice with enum dispatch. They can't they can only use the types that you've provided variants for um But dynamic dispatch is like static dispatch in that it allows any incoming type unlike enum dispatch Um, but it does so in a way where you don't get the monomorphization So that's sort of the slice that dynamic dispatch does The other thing that it does is dynamic dispatch can be um Can give branchless dispatch, which I don't think you can do with enum dispatch So with an enum dispatch um If you want to call a method on the inner thing You need to match on the enum discriminant before you dispatch Um, which means you have to branch in the cpu, which means you have to rely on the branch predictor Which is usually pretty good, but it's it's an indirection anyway with dynamic dispatch. There's still an indirection But it's a virtual function call, which means there's no branching You just load and jump or load and call and Usually that doesn't matter But in particularly hot loops That difference can end up making a big difference, especially if the branch predictor turns out to not Hit very often So those are the the first two that that um come to mind Yeah, you so as uh alice points out in chat too like enum dispatch is extremely strict orphan rules basically the only Types you can use when using enum dispatch are the ones that are declared in the crate That declares the function that is being called or dispatched through um So it's it's like not open to um any other implementations um Okay, I think i'm going to call it there three hours and 20 that seems about right There are a bunch more questions, but they're mostly low question votes I'll make sure to read through all of them. It's interesting to read through all of them. Um Okay, i'm also very glad that the stream worked all the way through like this is my first wayland stream on this computer from Norway and it seems like everything is just kind of worked, uh, which is amazing My internet connection is fast enough even though it's wi-fi. I'm streaming over wi-fi right now But that makes me happy Okay, I'm going to call it there Um, thank you for watching if if you're watching this video on demand then I'm guessing you haven't watched all of the questions because why would you you can just click to the ones that are interesting Um, but if you're watching live, thank you for watching through this whole thing Like there are 350 of you ish who have watched most of this and that's wild to me But hopefully you feel like I said something useful And I don't know when the next Normal stream is going to be like rust oriented stream is going to be But now that I have the setup, I know that it works and I'm starting to be settled in I'm hoping it won't take too long Thank you all for coming out and I'll see you next time. Bye folks