 All right, Horst. Hey, thanks so much for joining us in your own little private tour of Amsterdam in our nice little Audi. Can you tell us a little bit about who you are and what your background is? Yeah, sure. Yeah, I'm Horst Kramminkel. I work for Orteg Finance for over 10 years. Started just after graduation. And I'm a personal life. I'm a father of four sons. Oh, wow. So I don't have time for hobbies. Yeah, right. I only have three. And you're now the second person I've interviewed that has had four kids. And I always feel like I have a lot with three. Yeah, our youngest are twins. So I'm cheating a bit. How old's your oldest? Nine. Oh, okay. Nine, eight and almost three. Yeah, mine I think are a little bit easier now because mine are twenty, sixteen and fourteen. So they're pretty self-sufficient at this point. You get some free time back. I just have to make sure I make fun at work. There's not much time left. And at work I have a background in high performance computing. I actually graduated here in Amsterdam, but I didn't saw the city that much. So I was mostly in the university. And started different engineering roles at Oratec Finance. And what we do as a company is we design mathematical models to make better financial decisions. Right. And wrap that into software and ship that as a SaaS solutions. Oh, SaaS solutions. Yeah, yeah. And so in your environment is like a big kind of Kubernetes and OpenShift? Um, lately. Yeah, yeah, yeah. We're moving there. So I was responsible for setting that in motion indeed. While our company, I think for the last 15 years we acknowledge we are, it's a 40-year-old, what you would now call fintech. I think. Yep. And formally it was shipping software made by econometricians for econometricians. So we had an R&D department specifically focusing on mathematics and econometrics. Yeah. Seven years ago I convinced my boss who's now CTO that we should also research technology because the landscape was changing so fast. Right. And every tier right on the front end, the back end also, database technologies that was the era of the NoSQL. Yep. Yeah. Yeah. I let myself a part-time position next to my engineering activities in the lab. Trying to look for what's next. What's next, yeah. Yeah. And then I became responsible for all new to the company tech. And yeah, we had a lot of cloud technologies also given my HPC's interest. Cloud technologies in our lab. I think our first Kubernetes experiments were in 16, 016, something like that. Playing around with serverless and yeah, with the cloud. But you were doing serverless in 16, in 2016? Yeah. Yeah. Yeah. Just after the lab that came out, I had to supervise. I also supervised students, master students. And one of the professors was saying, yeah, you should check out. Just Lambda. Lambda. Yeah. And actually we did research on where is the breakeven point back then to make it cost efficient. Yeah. Because if your containers are utilizing or having a high CPU utilization, then it's cheaper to use regular container technologies. Right. So that's one example. We did many. And after I think just prior to the COVID crisis with the CTO office, we revised the whole enterprise technology strategy. Oh, okay. And that was for me the opportunity to get all these technologies were in my lab for a while. Right. Into what I call the factory. Yeah. And so together with my colleagues with the chapter leads, we wrote out this strategy. And yeah, cloud native. And event driven as well. I mean, because typically in FinTech, right, a lot of stuff is, you know, because I did, I was a consultant for a long time. A bunch of that was in FinTech or financial services. I actually did some work for Thompson speaking of software as a service about making good decisions. But, you know, a lot of those architectures have been enterprise service bus, you know, kind of event driven architectures for a long time. Was your environment like that? Or was it more kind of, you know, because in some ways modeling is more like a, you know, you put in the parameters right now comes an answer. Yeah. Yeah. So we don't have, we have event driven architectures at the moment, just because the, yeah, the cloud native stack allowed us to embrace these, these new paradigms more easier than in traditional ops. Right. On premise of raising these type of architectures are not supported. But in a nutshell, we build a lot of calculation engines. Right. Yeah. And some of these engines are used by banks, large Dutch banks use our models from their mobile app. Oh, interesting. Yeah. So then we forecast kind of the probability of reaching your investment goals if you're an individual investor. And then we need to cater for an answer like an API within two seconds. Oh, yeah. So that's number crunching, multi-threaded. Well, it's hard, it's even harder to do multi-threaded, like then typical because you're trying to put all the calcs together. Yeah. Right. At some point. Yeah. Yeah. Which is adds another layer of complexity. Yeah. Yeah. Yeah. But yeah. So then that's more, yeah, some people call it microservice like we call it microservice or services. We, we are not married to it to microservice. Yeah. I'm still like, you know, it's a service oriented architecture. It's a, the idea is like it's services, whether they're big or small services, like do we need that? Yeah. Yeah. I mean, I think the, what I have liked about the microservice buzz term kind of has been kind of a push towards like the Unix philosophy of like, you know, this thing does one thing well. Yeah. And the way you get stuff is by changing it together. Yeah. Yeah. Yeah. So that's been a big help. Yeah. Yeah. Yeah. But at the end of the day, it's still a service, right? Yeah. Yeah. And in terms of doing stuff well, like forecasting the balance sheet of a pension fund is a single responsibility thing, right? Right. But you can imagine we have 600 lines of code doing that. Yeah. Yeah. And these are, these are, so we, our product ranges from, from web, web products with GUI without GUI with only API, but also more and more traditional desktop product. Yeah. Yeah. So it's, it's, it's biometricians, fouricometricians. You can imagine there are not thousands of users. Right. Right. And there's no business case to, to kind of have that kind of scale. Yeah. To, to, to, to migrate thousands of screens in the desktop. Right. And there we, we, we, so we do believe in thin client FAT servers. So. I also worked on training desks. So. Oh yeah. Yeah. Yeah. Yeah. And of course we, we, we, we package that now in containers and where we first doing distributed computing on bare metal and on self-managed data centers. Our enterprise technology strategy stated we do managed overdue it yourself. So. Yeah. We stop maintaining data centers. Interesting. Yeah. Yeah. So the kind of containerization is also allowed a lot more of your kind of code reuse in different scenarios. Right. Where you probably had a lot more custom thing, like one thing that worked for desktop and one thing that worked for, I don't know, mobile. Right. And now with the containerization. Yeah. Yeah. You get a lot more usage, you know, shared usage. Yeah. Yeah. Yeah. Yeah. But what's fun is that you get these, these advantages where you don't expect them. For instance, now our CI CD architecture is, is almost the same for Java web as for the back end of, of, of desktop applications. Yeah. And that's way easier to maintain. Right. If everyone ships their stuff into container, as long as it's a container, all the management and the GitOps, all these, we can standardize a lot of these. Yeah. And not caring about what's inside. Right. Right. And, and, and formally we have like branches and operations that do, do Microsoft technologies or do, do Java technologies. Well, exactly. Especially if you're like shipping to a desktop, right? Then that means you, you know, you need every version of everything in your tool. Yeah. Yeah. Yeah. You know, and you need to make sure you don't conflict with local, you know, which, you know, in a kind of a server scenario, it's a lot simpler, right? Yeah. Yeah. And that looks like this and it's going to work this way. Yeah. You know, but with desktop applications, that's a whole another nightmare. And it's not like, you know, with, with those kinds of tools, it's not like you can have, you know, a real simplistic desktop application because it needs to still perform, right? It needs to be able to utilize that, you know, all the processing power or whatever of that workstation. Yeah. Yeah. But not if it's a fin desktop and all the computing is done elsewhere. And that's, that's where we were moving to. Yeah. Okay. Yeah. Yeah. Yeah. Yeah. It's funny. I think the, you know, the thing client was like the hot thing, I don't know, 15, 20 years ago? Yeah. And it just kind of never took really, there were some scenarios where it worked. But, you know, I think we're finally getting back to that because of mobile. Yeah. Yeah. Yeah. Yeah. I think you mentioned the event bus. Right. Right. That's also event driven. It's also very popular now and also very powerful. Done right. Yeah. Well, and it's fun. That one is it always cracks me up because I actually was doing event driven architectures using com on Windows 25 years ago because I'm old. But, you know, it's, it's just funny. Like, you know, we keep looping and there's, there's a great talk. I don't know if he's still doing it, but he used to do it. This talk at Osccon every year, which is basically like everything in computing was like invented by 1979. Yeah. And then he just kind of goes through all the modern stuff and he's like, oh yeah, this is just like the thing they did in the 60s. Yeah. You know, it's just really amusing. Yeah. But in all the range also in the database also first we had to decouple compute from storage and now we're bringing back the computer. Right. Bring it to the data because you can't move the data fast. Yeah. Yeah. It's really. Yeah. And then yesterday there was I attended a few talks in the Edgecon track where they mentioned that we will start shipping more and more software to devices. Yeah. So I just moved into a new building on the campus and we, you know, I was, I was actually giving a brief talk about kind of edge computing and what it is for the students and, you know, and part of my examples was like, yeah, in our, in that building alone. So yeah, 16 floors, but still it has, you know, trash cans, you know, everywhere, but every single trash can has scales in it that you can get real time data of the weight in that individual trash can. And one of the really cool things that Boston University has done with their contracts is that now if they hire a vendor to do, I don't know, scales and trash cans, I don't know. Yeah. Yeah. BU has as part of their contract that they can have the data back and that data be public. Ah. So we're basically generating like data science, you know, data sets for, you know, essentially students to work on based on our actual consumption. So it's, it's really kind of neat. It's one of the things I've got to look into over the summer is like, okay, you told me this exists. I don't actually get it, you know, but, you know, because there's often a big difference between, oh yeah, there's a data set for that and like actually having it on my laptop, you know. So, so what do you think is, you know, kind of from your perspective, what do you think this is going to be the next kind of big change? Is it going to be kind of getting all those things to like a cloud native scenario? Is that really the next major point or are you already looking at something else on the horizon that you think is going to impact your, you know, approach to these problems that, you know, Kubernetes or OpenShift or, you know, these, these tool chains kind of help enable or help you think about? Yeah, good question. I think first of all, from our context, we acknowledge that this transformation takes time. I think even Netflix took six or seven years to go from on-prem to the cloud. And what we see now is the technology is actually pretty easy at some point. Yeah, if you're certain. If you get far. Yeah, yeah, yeah. You fought your battles and sometimes it converges and you see engineers becoming better every time in embracing new technologies and platforms play a big role in, in, in, into broadcast these new technology to our engineers. The biggest challenge now is to, to change the humans. Like the, the, the organization aspect to, to these technology changes. I'm a strong, strong believer of Conway's law. Yeah. And that the systems you built reflect kind of the organization behind it. Yeah. How you communicate is an organization. Right. And as a company, we have four different units. Mm hmm. And so you have four different pieces of software. Four different pieces of software, but also catering for different markets. So these, these cultures, although we are headquartered in the Netherlands and then we do have a large organization workplace, but I see so many differences within the teams already. Yeah. Yeah. And, and so, yeah, so, so how are you going about that? There was actually, we did a talk in Detroit with Ford, did a panel talking about their transformation, like the same kind of scenario of like, how do you bring, you know, your, your swath of programmers to the cloud-native world? Yeah. Yeah. But yeah, that's the onboarding part, but also in the running part. Mm hmm. Right. Like you should, your delivery process changes. Even the, the, the, the salespeople, the, the consultants, the whole workforce is, is, is affected one way or the other. Oh, that's interesting. Yeah. Yeah. And so, so what are you doing to try to, you know, teach them all of that? Are you, you know, doing big camps? Are you doing, you know, publications? Are you, how do you, how do you bring that information to them? Yeah. So, so every time we onboard a new product line, we, we are hosting these, these college tours with tailored, tailored sessions for, for engineers, but also for the non-engineers. Yeah. And what we, we, we have a product line. We were now onboarding the third one. Uh huh. And, and we're teaching way differently. Like each time. You've learned how to do it better? We, we improved, but also the content is, is different. Yeah. We're now busy with onboarding a team that has a, also really release cycle for configuration. Oh, right. Typically in the financial sector, uh, you should, you can't release fast always. If you ship a SaaS solution to a bank, you cannot do that twice a week. Right. You can, the bank is not equipped to do that or most banks are not. Yeah. So what happens is you don't ship new code. You ship new configuration. Yeah. More often. Yes. That's way different than them. We now implemented kind of the GitOps, a GitOps stack. Uh huh. People are not familiar with code. Right. So you're also like having to train your customers as well about how to like work with your products as well. Also. Yeah. Yeah. Yeah. But, but also consultants who are responsible for shipping the right, right environments. Yeah. Uh huh. Yeah. It's so funny. I still remember, um, so way back in the day I did a consulting project for Fidelity. Yeah. Yeah. Yeah. So, um, it's a finance world. It's actually a pretty small world. Yeah. Curious what you consulted them. Yeah. So, uh, with, uh, with the project we worked on, we were, um, uh, we were looking for a way to basically kind of like change how authorization and authentication worked in the existing system. Yeah. Yeah. And we, you know, budgeted for like a three month project or whatever to do this work. And when we were working on it, I kind of saw this way we kind of, kind of slide it in. Um, and so we were done in like a month. Yeah. But we, we missed the release day by, by a day, two days or something. Oh really? So Fidelity essentially paid us to stay on board for another month until the next release window. Yeah. Yeah. Because it was, you know, because we would have disappeared into other projects, you know, or whatever. And it was more valuable to them to, you know, pay for us to be around. Yeah. Yeah. And to risk that. Yeah. Configuration update, right? Or that release update. Yeah. And I still think about that with, you know, talking about, you know, continuous integration, continuous deployment. Yeah. Yeah. And, you know, where some organizations, I also worked like in Pharmaceutical where, you know, if you, if you make a mistake and you drop a transaction on the ground, it was a million dollar fine. Yeah. So under some conditions. So, you know, there's some scenarios where you want the same quality of update and change and that kind of stuff. Yeah. But you don't want the same pace. Yeah. Yeah. And I think you see that in, you know, finance, like I said, you see it in Pharmaceuticals. Yeah. So it's interesting that you're experiencing that and trying to figure out how to, you know, how do you integrate that with these kind of modern development about, you know, methods. Yeah. And when it really comes to the surface is when you design your GitOps flow because that's really fun. Yeah. Actually last week we did that with the team that's responsible for configuration. Mm-hmm. And then that's where the technology and the process come together. Mm-hmm. Right? So then we thought, okay, how do you want it to work? Right. How does it go right now? And a lot of handshakes, a lot of handovers, which we now in the current realm we can automate. Right. And also with the GitOps it's so powerful. You can program your policies. So if someone doesn't update in configuration, you can say, okay, someone who's not the author, who has this role in the AD group of the company is allowed to merge this pull request. And also if someone releases configuration, how do you interact with the engineers? Because they are responsible in the end for the total surface. Right. That is shit. So that's a very nice, these are very nice whiteboard sessions. Yeah. Yeah. Where you kind of shape both your technology and process in one. Right. Yeah. I think I missed a turn. So I was just checking to see if I a small detour. Yeah. So tour this little part of the upstream. Yeah. So it's, I think, you know, as we, you know, as we are able to divorce a lot of the you know, kind of the operating environment from the application environment, you know, using containers and Kubernetes and, you know, things like that. I think it's, it's getting interesting about, you know, we're moving more and more to almost like, you know, a graphical design of the, of your systems, you know, where you have to like have kind of call outs and stuff or like, okay, this happens in this kind of you know, these kinds of things. I don't think we have the tooling anywhere near there, but it's really starting to feel more like, you know, you're, you're kind of writing a runbook for your operating environment in this very, you know, mostly yamble, you know, that, you know, and, and you kind of just say, okay, here's the runbook. Oh, computers now go follow the runbook. Yeah. Yeah. Luckily I have a colleague in the team who does a lot of withdrawal.io or whatever you have many, many places you can make, make these processes tangible. Right, right. By just using icon sets. Yep. These also come with Kubernetes icons, with your hyperscaler icons. Oh, I have to try that. We do that a lot. Yeah. Actually, because in the end it's what you say under the table. Yeah. We sometimes say, how do you like them? Yeah. It's a little throwback to what I live in Boston. Thank you. But yeah, I totally understand. Yeah. Yeah. So, I mean, you know, I do the same thing with a white board and my students all the time, right? Where I'm like, okay, you know, this is how these pieces fit together. And those pieces fit together. And my drawing is terrible. Yeah. This is my handwriting. So it's always this like chicken scratch all over the board. And my students want to take a picture of it so they can remember it. Yeah. Yeah. I'm actually going to go to the next one. And I'm like, you know, can you actually read this? Yeah. Because I can barely read it. I wrote it. Yeah. But so it's, yeah. But that visual component I think is really important to a lot of these scenarios. Yeah. Yeah. So a lot of stuff we do is it's about visioning the possible. Mm-hmm. And then Taylor, Taylor, the processes and technology. Yeah. Yeah. Yeah. So let me just convince it that we're going to where I want to go. I want to drive there. I think this will work. I think I make Google Maps very angry. Mm-hmm. But I'm not familiar with here. Yeah. I'm starting to get the hang of it. Yeah. But yeah. So what we want to do is actually just kind of take a right, right here. And then there's an entrance into the parking lot. Oh, nice. So. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. Yeah. So, well, thank you so much for your time. I really appreciate you joining us for a little ride around Amsterdam. You know, I hope you enjoyed yourself. Yeah, it was fun.