 Three, two, three. I have a fan club over here. Thank you. So thanks to all of you for coming. I really appreciate it. And if you're here, you're late. You fail. Fail. No, come, come, come. I'm just kidding. I'm just kidding. I'm not kicking you out. There's any seats somewhere? There's one seat if you want it. any other seats there's one here any other ones so if you want a couple seats there's some mixed in so thanks for coming on the Saturday morning so by being here you're already doing a fantastic job of leading yourselves by getting out of bed I didn't want to get out of bed this morning actually I woke up at six and went to the gym breakfast and all that stuff I didn't want to do any of it but I did it anyway because that's kind of part of it so this talk doesn't really have a title I think I kind of wanted what would we call the meetup it was how to win at something I don't remember right because we're all here to win right we're here to win so that's kind of the idea and the reason is that when I talk with people over the years sometimes people want to talk about technical things but when I do mentoring probably 80 plus percent of the time it's been on things like how do I negotiate my salary or how can I get a better job in data science or how can I become a manager or something to this effect so there seems to be some underserved need for talking about these kind of soft skill topics it's really really common I don't know it does anyone here who here is like a manager or leader a boss or something probably everybody right most of you somebody none of you I'm serious I really I really need to know this because it will help me it will help me structure the top so who here works somewhere where you have a boss yeah you have a boss somewhere Dimitri has no boss sees King the world okay cool everybody's got a boss I got one too I'm sure kind of well I might that's me but the same principle so I'm sure that when you have those discussions with your boss hopefully those kind of topics are coming up so do you talk with them about that do you ever ask them questions about how you can know nothing you do so Eugene's winning anyone else winning not really kind of sort of okay but it's a big topic and if you're not talking with your boss about it probably you're thinking about it and you're probably wondering about it and that's why you hear it anyway so some background before we get into it if you're on Twitter you can find me there if you think what I'm saying is awesome a Adriak hashtag leader if you think it's just a bunch of crap a Drake thought leader no one will know the difference but us so it's a nice little voting mechanism and this is not my idea by the way I took this from Josh wills oh you can come in come on join the party are there more out there get him in get him in really so if you agree with the stuff hashtag leader if you don't agree with stuff hashtag dot leader that's a little voting mechanism if you want it so background on me that's kind of me I've been into technology stuff for a long long time I liked math when I was a kid I like computers when I was a kid I like computers before I actually ever had a computer or saw a computer I don't know how that's possible but I actually have proof of this somewhere when I was a little kid making stuff in kindergarten so I've been into this stuff for a long time and when I was earlier in my career I used to think that tools could solve a lot of things and I would try to solve problems with tools and after a while I realized that tools are kind of off to the side the biggest difficulty that I faced over the years are actually people related problems how to solve problems between people how to get people to use the tools that I think are good how to get people to solve problems in a way that makes sense aligning people's priorities and incentives because if you have two people that want different things they're just gonna fight they're not gonna work together these are like the actual hard problems of making progress so it's not really about the technology but that's how I got started so that's kind of my my background in my perspective how many of you are currently in technology related roles where you either write code or do something with IT infrastructure or something sort of like that most how many are in kind of management or maybe product management or some other kind of role where you're dealing with a lot of people project management kind of stuff these kind of things but you're not really doing technical stuff most of the rest sorry how many of you are doing both sometimes some product managers really don't want to write code but some people who do coding stuff end up having to also do project management or product management it does happen so that's kind of the perspective that I have on this and over the years I sort of came to this conclusion that it's more about people and leadership than it is about technology and tools even though I still do tech stuff some of you maybe were at the talk I gave a month ago or two months ago so I still write code all the time because I like to and it's fun but that's not what I do for companies when I advise them and I do advise quite a lot of companies around the area and usually this is the issue so I talk to them for a while and then I say ah well it's a leadership problem it's almost always true because the people are usually smart the tools are usually fine but it's usually an issue of the company priorities are not clear people don't know what they're supposed to do or maybe they don't agree with what they've been told to do maybe they haven't been motivated to do something maybe they were just told to do so which is not a good idea most people are smart that's an important axiom if you don't accept that axiom that most people are smart in the sense that they either know what they're doing or can figure out what to do pretty quickly with your help if you don't accept that part then might have some tough time doing leadership stuff might have some tough time we'll get to why later so this is what I see most of the time it's I usually never go into a company I say oh well your database is just wrong if you just fixed your database then you would make a trillion dollars and your product would be fantastic and you would have all the users and then and then whatever you'd be bigger than Facebook or something like this that never happens never happens so it's not a technology or tools thing it's actually a people thing and leadership thing but we didn't actually talk about what leadership is yet and that's kind of important because if we're gonna talk about it we probably ought to define it otherwise we just sort of wasting time so I have a mathematics background in the first thing we do in math is we always define things so I I typically start with that so leadership is getting people to want to do what you want them to do that's one definition it's one commonly accepted definition there's a slight twist on this some people use it says leadership is getting people to do what you want them to do that's not the same that is definitely not the same there are a lot of ways to get people to do what you want them to do and most of those ways are not sustainable and they're not a good idea and oftentimes they're not very nice but if you get people to want to do what you want them to do that's actually a pretty good arrangement for everybody so this is kind of the definition I'm gonna go with the trick is how do you get someone to want to do what you want them to do the hard abrasive truth is you manipulate them and I say it like that on purpose I say it like that on purpose because when people say manipulation usually it has a really bad connotation right like you're trying to take advantage of someone or make the situation bad or better for yourself and bad for them or something like this right you're manipulating the situation but if you want to motivate someone regardless of your specific job if you're a leader in a company if you're a coach for an athletic team if you're like whatever if anyone has a personal trainer at the gym right they're they're trying to motivate you which means they're manipulating you into wanting to do something until you've actually done it enough times that then you want to do it as a habit and you become better for it so the idea is if you want to lead people you have to get them to want to do what you want them to do that's kind of the the overall framework and you have to do this in a positive way because you believe in them because you want to take care of them because you think they have potential and you want to help them find that potential which gets back to our axiom is everyone smart that was your question sir gentlemen that was your question right is everyone is everyone smart or do they just look smart if you're just looking at everyone saying oh you just look smart but you're not smart you're gonna have a hard time with this one because there you're really gonna have a tough time getting people to want to be more than they are because they're not gonna believe it so yes I think everyone is is pretty smart so it's kind of a big deal actually because when this fails yes smart people are usually more humble so it's usually not a problem so arrogant what I've I'm sure I'm sure it is sometimes but in my experience usually people who are genuinely smart kind of question themselves a little bit so they're not so arrogant and you can raise the Dunning Kruger effects flag or whatever you want to do but it just in my experience people who are who are really smart also are kind of critical about things including themselves so it's usually not an issue and arrogance is something that we'll we'll get to in a minute it's but that's an important topic so if the leadership fails then everything fails so without raising your hands because I know some people might be embarrassed about this but I would imagine a lot of you are kind of unhappy with either your boss or your boss's boss or something like this because they don't help you they don't grow you the leadership's not great it's confusing they manage you too much they manage you too little some something like this it's very very common and it's most of the reason why companies have trouble it's most of the reason why companies have trouble so when I'm advising companies 90% of the time it's it's something about leadership or operations that's causing their problem very very rarely is it something that's technical and even if there is some technical problem that's not causing the whole company to kind of crash and stop it may be an issue with one part of one product or something like that but it's not something that just spans the entire company so it's a big deal so the main thing that I wanted to get into today that I wanted to start with is I saw some questions you posted the form where people could post some questions right so some of you maybe or some people who are not here I don't know some mix submitted some questions and a lot of the questions were what can I do to become a leader how can I get a promotion what do leadership roles in data science look like these kind of things and the first answer to that is if you're going to lead other people you kind of can't do that effectively until you can lead yourself effectively yeah you can move around you don't have to you don't have to hide you don't have to hide yeah there's some chess here you want some chess everybody comfortable cool so before you can lead other people you have to lead yourself and if you can't lead yourself effectively no one's gonna follow you that's just kind of the way it goes you have to lead by example when you are a leader so if you can't even take care of your own affairs then you're not going to be able to help other people take care of theirs that's just a fact so before you start thinking about what can I do to get a leadership position in my company what can I do to get a better job or a better title or something like this the first question is are you actually leading yourself effectively and the answer should be not really I think of this all the time so I'm constantly working on leading myself more effectively all the time what can I do to improve what can I do to be better do I need to get up earlier do I need to stay up later do I need to work harder or smarter do I need to work on this myself or do I need to delegate it or outsource it you know all these different things I'm doing this for myself all the time and if you don't start there why would anyone follow you in the first place so step one kind of look in the mirror a little bit before you think of what you should do now then once I tell people that they say well I don't feel confident to be a leader actually I don't I don't quite know what I'm doing I don't feel confident about this how can I build confidence which is an interesting question it really is and the answer is you don't really need to you can just kind of fake it till you make it and it sounds kind of stupid but it's true it really is so as an example what's the difference between someone who's brave and someone who pretends to be brave is there any difference there's no difference what sorry Alan Turing is the difference Alan Turing is the difference the Turing test if you behave like something you are that thing in the end who is who else and you know what else can you say right so what's the difference there's no difference you might look smart you might look smart but you're not smart so you're saying you pretend to be smart okay I think we're seeing a pattern here I think this definitely okay so some people are pretend to be smart but they're not that smart how long does it take for everyone else to figure that out five minutes wow really I can't I can't figure out that fast we should hang out more I can learn a lot okay but in all seriousness there's no difference between a brave person and someone who pretends to be a brave person it's the same thing so when the topic of being more effective as a leader even though no one reports to you because just to reiterate didn't say again like leadership doesn't mean people report to you I probably should have said at the beginning so if anyone thought that you're not a leader if people don't report to you no no that's it's not like that at all so leadership is influence that's all it is leadership is influence and how you influence people has very little to do with whether or not they actually report to you and the higher you go on a leadership hierarchy in an organization oftentimes the fewer people report directly to you and you have to do more work influencing other areas of the organization so it's a good idea to start with that in mind leadership is influence it's not who reports to you it's not how many report to you it's not it at all so if you're wondering how to start being a leader you can fake it till you make it try to do a good job start with yourself first and then by being able to lead yourself more effectively you'll project that out to others but how do you kind of do that in practice so a lot of people just think that they're gonna be like a manager or a boss and I hear this a lot in Singapore I want to be a manager no I don't want to be a manager that's not what I like to do it's I'm not a fan of management stuff I'll do it because it needs to be done but it's not my strong area it's not what I enjoy doing I don't like doing budgets I don't like doing project plans and headcount estimations and I don't like doing all that kind of management stuff I like doing more of the people stuff it's more effective I found what I see in Singapore a lot is people say they want to be a manager I say why do you want to be a manager and I say well then I can sit in my office and tell people what to do I've literally had people tell me that I'm not making it up I'm not exaggerating they've literally said that I can sit in my office and tell people what to do and I can sit on my computer and they can do all the work that's that's what they said so I'm not Singaporean I haven't been here that long I don't have a an opinion on the matter that's just what I heard so it's kind of this approach like oh well if I'm a leader then I can tell people what to do no because leadership is getting people to want to do what you want them to do it's not telling them what to do those are really different so you can't go about it this way that's not you can't just start and say well I'm the boss do what I say so what do you do you need these two qualities if I had to pick like two qualities for people to work on for leadership and I work on these every day every day I work on these qualities you have to be humble have humility you have to know that you can learn from other people that other people are smarter than you that if you're leading other people they should know more than you about their specific domain and that shouldn't scare you if it scares you that people know more than you you won't give them the freedom to operate then you're gonna micromanage them and you're probably gonna do a bad job because they know more than you anyway so I think one of the questions was what do you do when you know more than your boss about data science stuff something to that effect and the answer is maybe the boss needs a little more humility so that they can have some trust in you and let you run free so humility is a big one you got to be able to learn from other people doesn't matter where they are it doesn't matter where they come from I was talking to a company that has a an innovation center in Singapore which is like the cool thing to do now a lot of these big companies have it they open innovation centers thank you EDB and it's it's really appreciated but you work at the innovation center somewhere which one which one that's the ages so there are a lot of them right even we have people in the audience working from so I was talking to people at the innovation center the other day and they were actually having kind of a disagreement internally in their center about who can have good ideas which is hilarious when you fact when you figure they're an innovation center like this is and this is what I mean the leadership thing is real it's a real thing like you're you're talking to the best of the best people from this big mega corp that has set up this innovation center in Singapore spent millions of dollars to do it and they're sitting around arguing over how senior you have to be before your ideas of can be good it's a real thing and the argument was can someone who's like a custodian or an assistant or something have a good idea obviously yes obviously yes I've had in my personal experience cases where people from any level of any organization can have a fantastic idea but if the leaders don't have the humility to be able to accept that maybe their ideas aren't the best that really slows everything down discipline is the other one you have to be able to do things that you don't want to do that need to be done that maybe aren't fun you have to be able to do that every day and hold yourself to high standards and all that stuff all the time and that doesn't mean you need to be strict and and angry at people and things like that no that's not what it's about it's about being reliable and responsible it's about taking ownership of things even if you don't fully control it because that puts you in a mindset where you can try to influence them and leadership is influence so they go these are kind of the four main points that I try to get people to work on and I work with people on these points so if you're actually in a role right now where people are reporting to you or if you're just trying to lead yourself these are kind of the things to focus on these are the things people need to understand when I go to a company I ask these questions to every person that I talk and most of the time nobody knows the answer to these questions so if you if you look at think just think for yourself wherever you work do you know what the goal or the mission is for your company and I don't mean like the one that HR puts on the job descriptions I mean like the actual real mission that you have because those are not always the same right why does it matter that's the big one that's why it's in capitals because it's important why does this goal matter how are you individually contributing to that goal so there may be some really big at the executive or the board level they come up with some goal right hopefully it's a good one and they tell you why it's important because they're fantastic leaders and they want you to understand why what you're doing matters and then as that trickles down through the hierarchy of the organization the different levels of leadership help you understand why your work that you are doing today influences that goal and of course they have to tell you the constraints why do the constraints matter because if they don't tell you the constraints they have to tell you what to do every single thing and that's not the idea you want to tell people where we need to go and why it matters give them the constraints and then let them figure out how to do it that's the balance that's the balance so if you don't take anything else away from this whatever it is 25-ish minute talk please please remember the last slide humility discipline and these that's that's it if you get those things you'll be better as a leader than probably 95% of the people that I ever deal with and you laughed at some of the questions I said that I've been asked before you'd be surprised at some of the other ones I didn't mention so these things are really important and I never never talk with a company it's super rare that I talk with a company where everyone gets this what I do find sometimes is a department and the department might have a really good leader and they've crafted out answers to these questions for their people that happens that can be a happy department so you might have a engineering group or data science group or product management group or whatever and they all know the answers to these questions but that's only because the head of product or whatever has crafted that they're completely detached from the rest of the company but they're having a good time so they think it's okay mission statements are tricky they really are because they're hidden miss right they're really short and sometimes mission statements get crafted so that people can take them to mean whatever they want them to mean that's like a political mission state some people will talk about smart goals and all this other stuff for me what I've seen work really well in companies I'll give an example there's a so I used to do work before in all my travel industry some years ago and there's a company in Indonesia called travel okay I don't know if any of you know travel okay they they have a mission of enabling mobility that's their mission enabling mobility and when I first heard that I was like that's kind of it's kind of broad I don't know I don't know about that but then after I thought about it for a while actually it matters quite a lot there are a lot of things that change once you enable mobility in a society people can get better medical care economic opportunities increase people can spend time with their families there are all kinds of things and ways in which society benefits when people are more mobile this is you can read the literature on this this is pretty well documented so after I thought about that that actually that's a really good mission the issue is that even if you have a good mission like that but you don't explain it to people in the right context they're not going to get it right I didn't get it when I first heard it I was like enabling mobility sounds kind of weak sauce I don't know but then after I thought about it for a minute that's there actually a lot of important aspects to that so something like that where you have a mission that short that's compact that people can keep in their mind all the time if someone has to go to like a knowledge base or a wiki or something to remember what the mission is it's a bad mission that that that I can tell you for sure for sure which is why these questions are important the answers to these questions should be short enough that people can keep them in their mind at all times this is not like where you make a big five-page document and you put it on your knowledge base no because people are going to make a thousand little decisions a day and they should all be guided by the answers to these questions and it's the same with the mission so so that's over you if you want to start on the path to being more active in leadership capacity my advice take it leave it start with yourself become someone who you would follow as a leader take the actions that you need to take to do that that doesn't mean you have to want to take the actions that goes back to the discipline right if you need if you know you need to do something you don't have to want to do it in order to do it I didn't want to get out of bed this morning honestly I really didn't I really didn't but I did anyway probably a lot of you were like oh man I'm tired Saturday up why am I gonna go to SMU at 10 a.m. on a Saturday to go to a data science meetup why did I even RSVP for this thing but you came anyway that's a good show of this most of you were on time very good show of discipline those of you who weren't you know who you are I'm not shaming thank you do you want to do questions or do you want to go handle the questions in the past sorry process or person persons good leadership environment starts from good high leave it over good high it's important to find proper process something like that so my question is you think that everyone you can be a good leader this everyone in every environment or it's important to have good environment inside okay so the question is basically what matters more the leadership or the team is that kind of what you're getting at yeah okay in percentage so you had a couple of other points embedded inside your question so yeah yeah yeah so I will first would like to go to those points and then we can actually go to the end question so one of the points and the one that I think is really relevant is recruiting and recruiting bringing people into an organization so that's something that I take really seriously because people don't need to be the absolute smartest and they don't need to be the absolute expert in the world in whatever field it is but they have to be able to work well with us there may have been a time when and it was true when I got into technology stuff in the 90s where you could be a tech person and sit in a machine and just code and that was it that you didn't have to really talk with anyone and those times mostly are over so when it comes to recruiting you really have to make sure not so much that people are technical superstars because if they're pretty smart because most people are pretty smart right then they can learn they can learn right and I'm sure people in this room have had that situation where they've submitted their CV to some job and they said oh well I mean we're looking for someone who has experience with Java 8 and you only have experience with Java 7 so clearly we can't hire you or whatever stupid stuff right this happens all the time and it makes no sense right so on the technical thing smart people can they can figure it out that's not a big deal but what takes longer is interpersonal skills it takes a long time to develop interpersonal skills and so that's one of the things that I look at when I'm hiring people curiosity discipline interpersonal skills those are the three main things if they have those they can learn whatever else so that doesn't matter so that kind of touches you're recruiting on the inbound side now then the other side to that is what happens when you end up with someone on your team and they're just not fitting they don't have the right I shouldn't say the right mix and I don't like to use the word culture fit because that's like a catch-all but some people are someone's asking about arrogance earlier right there they're yes it's there was some there were some people there are some people that are more interested in advancing themselves than advancing the team they're more interested in advancing themselves than the team that's a problem so when I go into any conversation I'm not here to be right I'm here with that's different it's really really different if you go into any interaction to be right you might have a hard time you get it and there are people who are like that they go into these interactions they say I want to be right I want to be the top I want to be the best and those people come out really quick and you have to deal with them really quick so obviously help them try to adjust their perceptions of themselves and others try to work with them but if they're just not getting it you have to cut them out fast because it becomes really toxic to people on the team so that's kind of the recruiting and the retention part now then the other part and kind of the meat of the question was what matters more the team or the leader right I don't know with a percentage so I can't really say but there are many examples and I've been part of them where you can have a great team and then you swap out the leaders and then the team starts to degrade or on the other side you can have a team that's maybe not so great put in some really good leadership and the team starts to perform they start finishing projects they start shipping stuff they start taking their work more seriously and taking more pride in it so what I've found is the best reflection of whether or not a leader is effective as if their team is effective and I don't mean effective for a little while I mean on an ongoing basis so I would say that leadership matters more than the people in the team and it's been the same for me if I've been in a if I've been in a team and the leadership is bad it's hard for us as a team to perform it's hard for me as a team member to be able to perform because I may not know exactly what I'm supposed to do I may not know clearly how all of the people in my team are supposed to interact so that we can make progress because the leadership is not as effective as it could be now I take responsibility for that because there were those questions which so going back to those questions so with these questions these are the things that you should be answering for people as a leader but they're also things that your leadership should be answering for you and if these questions aren't answered for you then you have to take the responsibility and ownership and figure out what the answers are and usually you can kind of find a way to meet in the middle but if the leadership is not that effective it makes it really hard to perform as a team I've had that experience myself where I try and try and try and try but the leadership's not it's not really that great so I would say the leadership is more important than the team your slides are great and very motivating but it looks to me that you are not talking with one hard part is the conflict management and when you lead yourself hopefully you don't have that much conflict I do I have conflict with myself all the time but does that conflict resolution that you apply to yourself applies to when you are leading a team I can't be as strict with other people as I am with myself that's the biggest difference that I would say so with myself I'm quite strict so when there are conflicts between people in the team if I'm leading the team then I take that as my fault that's my fault for what it doesn't matter the specific reason why there's a conflict it's my fault I take full responsibility for it always if it was a bad hiring decision if it's I don't know somebody's arguing because of it doesn't matter it's always my fault so I try to figure out where I made a mistake oftentimes it's because I didn't communicate one of these things well enough maybe I didn't communicate some constraints or some direction or something like that as effectively as I could that's a big one but for me the first step in conflict management is taking responsibility for everything and then doing what I can to fix it because when you deal with people and there's conflict people have a hard time separating their emotional reaction from their logical reaction it's very normal and so the first thing that you can do to help that is take full responsibility for everything that changes their emotional reaction from defensive to well it's not all your fault it's not all your fault right and it changes the dynamic really a lot so I don't know if you're talking about conflict that you have with someone else or conflict that you have as a leader with someone on the team or conflict between two team members and your leader because the goal is to motivate them yeah yeah it's usually not a good strategy yeah so what I do when I'm in that situation and there's a similar question someone asked about about bosses I think that we're kind of like micromanaging and how do you deal with this and it's kind of similar because there's always conflict there right and what I try to do is to become very useful to that person I become very useful so first step except responsibility for everything doesn't matter what it is one that changes the perspective of the other person because then suddenly they want to say well it's not all your fault it's okay it's okay we can work on this together and it changes things the second thing is it changes your mental perspective and it gets you more focused on the things you can control and that you can influence so that's what I do then the second thing is I try to be useful to people and that's important because you have to completely set aside your own ego so if anyone has ever thought well I don't know if I want to do this for this person because they always take the credit not a great idea not a great idea you're really getting in your own way your ego is preventing you from actually delivering something useful if you're worried about the credit you're gonna have a hard time as a leader because the team always gets to credit the leader gets to blame that's the rule every time and if you're going to lead people you have to be willing to accept that so I usually go for those two things one just immediately turn it around and take responsibility for everything and then the that usually helps actually quite a lot and then the second is to try to help those people look better and usually that that will reverse it now there's always the case of the person who's kind of pathologically arrogant or whatever they just no matter what you do they're always gonna be a problem and that's something you kind of have to talk to your leadership about and your leadership should notice that pretty quickly and deal with that pretty quickly because it's usually very obvious it's usually very obvious when a person is just a problem and they're not getting it and they have to go so unfortunately it does happen sometimes but it's rare did that help sort of sorry we had to raise stand okay the tools you have that can help you to improve the leadership you know one of the experience you have maybe from a previous experience that we can share with us you know exactly what you learn from an experience like that you know contextualize you know your theory or your perspective today so software management and tools do you mean like specific project management as if you have a project you know so you're talking from a project management point of you a leader in a project management team okay so the thing I see most common as an issue in that situation is unmanage expectations that's it that's really it because the business doesn't understand the difference between something that's easy to think about and something that's easy to build for them it's the same right so if you have a marketing department let's say and they want some kind of analytics product and they just think it up maybe it takes them five minutes to think about it and so then in their mind it was easy to think about therefore it must be easy to build which we all know is totally wrong it's 100% false there's no there's not necessarily any link between how intuitive or easy something is to think about and how easy it is to actually build and deploy so I spend a lot of time on that specific thing managing expectations and stretching out the timelines so that people understand what that does is it frees up the team so then when I go to the team they have more room to maneuver they have more freedom to operate right so that balance is really important do you have tools or to improve let's say project software development when you say tools or frameworks do you mean like software tools yeah no that's to me totally relevant I can't say it more plainly than that like that you can use a pen and paper and be an awesome leader or you can use a really fancy project management suite and have a really hard time it because it's 100% about the relationships and the psychology of the people that the tools are not that big of a deal I've gone through projects there was one there was one very nasty one that I worked on some years ago that had been already ongoing for like five years it was supposed to take six months six and it was still going five years later and they couldn't stop it because they spent so much that if they stopped it then they would lose all the money which is I'm not going to go into that logic but this was the situation right we ran that project with Post-it notes on a wall and we finished it in nine months I didn't use any spreadsheet I didn't use any project management tools I didn't use any anything like that I use Post-its on the wall that was that was how you so I'm not tools whatever works it's it's the same thing with programming languages I'll use whatever programming language works there are some I like more than others but that's not my primary focus my primary focus is not which one do I like more my primary focus is which one works the best because I'm not here to be right I'm here to win that's the difference oh yeah you can do that with Jira I mean there's all kinds of stuff yeah yeah but the thing is it's easy to get distracted talking about tools and tools don't do work tools help people do work well that's 75 and it was distributed across different locations so it's what I found is anytime someone is talking about tools or if I'm thinking about tools I always take a step back and I say okay did I actually am I actually solving the problem here or am I just do I just want to feel like I did and anyone here Java developer okay I'm only gonna partially install so the old-school javas not not so much in the new days but like 10 10 plus years ago there was there was this concept that people liked Java because you could feel really productive because you wrote so much code and there's there there's some truth to it there's a psychological effect from having written a lot of code right and and as with Java they thinking was well you kind of get that by nature of using it as a tool so people like it because they feel productive when they use it but if they use something like pearl for example and we won't get into pearl one-liners but you can do really short stuff in pearl so what's the what's the distinction there and there is an effect of okay I'm using this tool I'm doing this stuff I've got charts I've got graphs I've got metrics and that's a management thing that's a project management thing like you're saying like what are the metrics what are the how many hours do people spend on whatever it is how many hours do people spend on meeting but I would much rather just go to someone and say do you feel like you're spending too much time on meetings like that will have a totally different psychological effect so I think that's not a bad thing to do stuff from somewhere but you need to know the fact that how to address that because if you don't even record knowing those people are spending you know maybe not effective time in something stuff which is probably don't even know where you're going to start just to see what the team are moving towards but which exactly the part is not effective I think that still makes I'm not saying it's bad to know but in my experience there's there's people say a lot of laws con wise law and all this stuff but there's one that I actually like which is good hearts law just once you start measuring something once a metric becomes a target it's not a good metric anymore and so sorry please so please so when you if you start having meetings for example meeting time as a as a target reduced meeting time I guarantee you your people reduce meeting time I 100% guarantee it but that doesn't mean they'll be more productive harm and perform my revenue you know which product you which was not so much time just crunching the data figuring it out to get the management to make that decision do the right things something having the right tools that you definitely have the decision makers which is my content I'm not saying oh the goal is so yeah we see the meeting too much that's just all going to reduce the safety I think you to agree talk about this match expectations tools can be a way to bring these out in the open so I don't think that tools necessarily can tell you what the problem is I think tools often hide the problem that's why I'm not a big fan of tools I just ask people it's really that simple I mean this is I can't emphasize this enough dealing doing lead being a leader is about people 100% it's about caring for people making sure that they can grow that they feel safe that they know that you care about them and you want to develop that's it and anything beyond that is an implementation detail in my opinion so you might need it for some things but it's not critical so when it comes to things like like tooling yeah use what works if you if you need some bi tooling to figure out some things definitely use it if you need project management tooling because you need to have some kind of reporting for your finance department that's fine but that doesn't mean and I'm sure some of you experience this like you get time tracking pushed down your throat because some finance person wants to know how much of your hours they can build to some other department it's it's so demotivating because you feel absolutely like some cog in the machine and they want you to like log all your hours for everything like I was in a meeting for 25 minutes and then I was writing code for 35 minutes and it's crazy right and so what I what I've always tried to do is to separate that so the team wants to work on work stuff they don't want to work on logging their hours so I try to find ways with the finance department for example where we can satisfy their accounting requirements without having to make the team do the logging of the hours or something like this break apart yeah see but I'm sure the SAP sales people made a nice commission so it's quite okay you take one more question before we start panel because we're going to be late okay your hand was up first agreed last question okay so in an ideal situation there's going to be some big goal and it's going to responsibility for different areas of that are going to be split and then split and then split and then split and so first you have to know let's say you're part of a team right as a team we have some kind of goal that we have to achieve and we're gonna have some kind of boundaries or constraints on the goal it may be technologies we can use could be could be time that we have could be scope can't be time and scope bound that's not allowed that's contradictory you can't do that but there there are constraints like this that come up now as a leader if you're briefing your team on this topic they need to understand the constraints this is really important because if they understand your intention why the thing is important and what your overall goal is and they understand what the constraints are they don't have to come back to you all the time to ask questions that accomplishes two things one it frees you up as a leader to be able to keep some of the other stuff away from the team like the time tracking right but also it gives the team a much bigger sense of autonomy and freedom because it's up to them how we're gonna accomplish this that's a good question you're the expert how would you accomplish it I mean if you if you need something I can give you advice I can give you support if you need more money if you need more people I can try to make that happen but I'm not the expert just because I'm the leader doesn't mean I'm the expert right those those aren't necessarily the same thing so the constraints are all the things your team needs to know so that they can operate semi-autonomously and it's gonna vary from team to team and project to project but that's that's the the reasoning for having those constraints included you want to be able to give a brief to the team say this is our goal this is why it matters this is my intention for when the goal is accomplished and these are the constraints on accomplishing it so that's all the where do we want to go and why does it matter and then all that how do we get there is for the team to figure out on its own yeah right right and the the trap that a lot of people fall into is they go way overboard on the constraints they they go overboard on the constraints too many constraints most of which are unnecessary so the important thing with the constraints is that they have to be only the necessary constraints so that you're maximizing the autonomy of the team so that's that's a really important part that's also business constraints so for example take for example you will get an e-commerce company you have a lot of new products that are on your platform every day so maybe one of the business goals is to introduce new products on your search page on your catalog page and the business says okay help these new sellers generate sales help help them grow and the constraint is overall conversion doesn't drop by more than 5% for example so that's a country do whatever it takes but overall country doesn't drop help them so that's a country simple this or one single country so I mean imagine how easy it is if you're in a team you say something like that we need to develop new system to this and get more sales for the customers it has to be done in this and this technologies you have to interface with this and this team and the constraint is no matter what you do conversion can't drop by 5% however you choose to solve it after that is up to you as a team and that makes it really clear what you're supposed to do and really clear the parameters in which you have to operate does that help okay let's go for a five minute toilet break and thank you Adam for this section once again thank you for coming down this early morning so before we start maybe everyone knows a lot about Adam but well maybe cool may give an introduction for myself okay hi so I think some of you may have seen me before at at the meetups and all these so I'm actually you one of the co-founder together with icing on this middle group so thank you very much for the support so far I really see a lot of people now the community has been growing that's not part of myself intro but yeah so I'm I'm actually in the field for quite a while I think some of you may have known me and based on the read-up that you have seen right I also do teaching at some of the ISL's over here but for now I'm actually freelancing now as a consultant and as a trainer more of training my passion besides the helping businesses to learn more about their science and all these it's also helping more people to understand the field to actually help get more passionate people into the field as well so that's my self intro okay thank you okay so we have 25 questions from based on the responses and I can sort of see that there are two groups of these kind of questions one kind of question is like oh how do I get a data science so general data science questions and then the other one is more about leadership in data science or managing leaders in data science so we'll do both but I'll just like to get a few from the audience which which ones which set of questions are you guys more interested in general data science please raise your hand so I can get a vote and then leadership in data science I think okay it's about the same but I think general data science sort of a dress up a bit so you can jump in anytime as well yeah that's right yeah so I guess one question that everyone asks is how does one who is a mid-career transition into data science from technical background maybe sorry okay let me read this question I have some SQL developer skills what what's all this machine learning about and how can I get into it so so maybe I'll just add it on so how did you pick up your data science skills are there daily rituals or habits I'm sure Adam has a lot to share about that what's your advice from someone from another field who wants to transition to data science yep so as a base question I get a lot is do I need to study computer science or math to do data science and the answer is 100% absolutely not so I've worked with fantastic data scientists and my favorite example from a different field was someone who had a PhD in Russian literature and they weren't Russian so just to be really clear about that they were they were from the I think they were from the UK but they get a PhD in Russian literature and they became data scientists and they were really awesome they were really awesome and the reason was because they understood that they needed a balance of technical skill communication abilities business acumen and understanding that it's really quite a wide array of things the technology is a part of it and it's maybe the hardest part for a lot of people to pick up but it's certainly not the case that you should only have studied computer science or math or something and in fact for many people I would say it's easier to pick up the technical skills than it is to pick up the business and the interpersonal skills so it's case by case of course but it's definitely not a requirement okay I started off as a I'm a columnist by training actually my first degree is actually cons what I can share with you is how I actually move into the field rather I think that that could be something for your reference I started off as a con student I graduated around 2002 2002 2003 unfortunately what happened was that during that time there was a SARS period in in Singapore I could not find any job as an economist so I actually was I had a year blank so by a blank I can't find a proper job then but what happened was I stumbled into education research I did education research I did for a while and stumbled onto this software called SAS and that's how we are going to stay at this guy using SAS okay anyway so having said that but then I slowly moved on and transitioned in into the field of doing analysis work so from the education research right then move on to actually in the banks as a as a risk manager as a risk analyst I did a lot of modeling a lot of analysis and all this so I might I will see that the whole thing that links up right is actually the analysis queue the technical skills that helps helps me to actually transition into into this particular field altogether so so I think going back to the question of let's say if I'm a mid career I do see a lot of careers want to change job into moving into data science or even to data analyst I feel more importantly is you need to find an anchor point and from this anchor point right you move to take the next step you create some small means for yourself first like hey yeah this is something I can do maybe this is something I I would like to develop further and from then you take bigger steps and you also take you also learn more along the way I think learning having a strong hunger for knowledge and skills is actually a very is a is a plus point for all data scientists first there's no way we can learn everything but we want to learn everything best the conflict most data scientists will have actually I would say yeah I think who's response is great let's try to spend a bit more time to try to dissect it so cool if you think about it he was mid career in economies himself then he moved to risk management what do you do with what technical skills did you need to do risk management in a bank okay I think my econ's background actually helped me to transition into a into a bank but the second thing was I have to say it was learning says learning says actually got me the interview so knowing having the econ's background itself is not enough but knowing that particular tool that a bank use help you get a foot in the door in the interview yes okay yeah so just to add something to it I was mentioning that you can have great data sense and do real design we both having a big background but in practice in most companies that they will interview in interview you on first our technical very specific thing you know start we come with programming interviews and stuff to bear train that you actually want to get to go out to SAS was a way but most of them you can't land the data science job with those cases really to check that's if we if we were to just because I get this question a lot when I do mentoring so I also did mentoring like with data science retreat in Berlin which is kind of like a three-month program where people polish their their background so that they can actually be productive data scientists and the question that comes up a lot is what do I need to learn to become a data scientist then you also mentioned people who have a technical background they want to make the transition and people who already have some technical capability are at an advantage in the beginning and the reason is for exactly what we're discussing when the companies are looking for what to hire they're going to look for something concrete so if you have a technical background it's not so hard for you to do some projects put them on github or whatever now you have a portfolio and people actually can see that you could be productive technologically and if you're talking about what you need to be productive as a data scientist you have to be a pretty good software person you have to be able to write code like that's very difficult to be productive in a data science team without being able to write code so if you if you haven't gotten to that point yet then that's where the discipline comes in you're gonna have to be disciplined about it and that's probably why you asked me about habits because Eugene knows that I'm big on habits just as an example I like Haskell which is a functional programming language I didn't know it years ago I decided I wanted to learn it so I started getting up at five every morning so I could spend an hour every morning learning Haskell that's what I did I did it for some months and then at the end of that then I was comfortable writing code in Haskell so it's strictly a discipline thing and people say well I don't have enough time I'm busy if something's important you make time and there are things that you will choose not to do it's very rare that I find someone who literally has no time most people can get up an hour earlier and they just don't want to that's a different problem that's a discipline problem it's not a time problem because I just happened to see what we do in the YouTube there's a you can search it's called data science superpower so there is a guy who's really talking about his experience from a single person to data science and what the language you need to learn what exactly the things you need to do and that's a very good sorry what's it called it's called data science superpower data science superpower data science superpower yeah okay the cool part is that all this stuff is on the internet right so you can you can pretty much learn everything really I'm just seeing this a very good stopping point because he was having a secret background all the way talking about the you know the two different type of person what the language means to each other and then he gave a real more examples of how he actually be a single person that create a data science program algorithm and so step by step share the code and scheme up and so that's a very good start to come but also I have some point what I agree with Adam is it's not regarding what you know but also regarding what you have done so I also have some past experience with Harrison people doing data science so what he is doing is not saying how many powers you have to have just most regarding what you really have been done so yeah that's a good point so if anyone's doing Kaggle stuff after the competitions are over I think you're totally fine to release your code and that doesn't violate any of the terms if I remember correctly so if you're doing this kind of stuff write it up do a blog post put the code on github and it's really makes a big difference the challenge I've seen with most people who are trying to get into data science is they don't document and write up their work and they don't put it out there and if you're doing that you're automatically ahead of almost everyone else because most people they'll do a Kaggle competition and then they'll get sidetracked and do something else they won't take the time to do that last five percent write up everything they did and put it out there and that really makes a large difference to an employer because one they can find it and they can start to get an impression of your work capabilities but also they know that you can communicate effectively which is really important for people who are getting into data science work and they know you can finish stuff starting things is really easy finishing things are hard and so if you're a person that can finish things and you can communicate and you have some technical then everything else is just a side detail the lady who is in economics or whoever else is in economics starts going to data science or closer to data science than you think with all your econometrics classes your modeling if you go into one of those Kaggle competitions and you try to apply your model then you work and you will by just reading the form learn about your modeling techniques and then you will be on the technical side quite well versed already I mean some particular conversion techniques and whatever you think that you are not talking about things but that are conceptually quite understandable for you so don't be afraid you are in something that doesn't sound like data science or any closer than you think today as well I think I feel like I know it I know that my skills will work in this industry but my CV doesn't look right when a recruiter looks at the TV and thinks it was an interview okay that's a recruiter problem that's really all it is and if you're able to get your CV to the hiring manager that is the data science lead who really knows what matters and not pull up points then it shouldn't be a problem I would I would completely agree with that and be very hesitant on recruiters you'll probably so this is like more of a career kind of process question but usually I don't advise people the only people that tend to have good luck with recruiters are people who are freelancing on specific technology or framework and they want to do billable hours on that framework maybe on a part-time basis for anyone else I I don't usually recommend to go with recruiters there are a few recruiters out there that handle high-level kind of executive roles and they are really good but they've been doing that for 10 20 years and they're very specialized in that the people who are recruiting for other roles they're just looking for check boxes and to them economics anything anything econ is is going to be oh well it doesn't say data science so that's that's their world and they're processing tons of resumes that's fine so it's it's much better to just have what I try to do and I just think it's interesting is I meet with people at different companies just to understand what kind of problems they have and invariably just by doing that you'll meet some of these hiring managers and they'll understand your background one of the best hires I ever made was this person and they had almost no technical experience in the things that we were actually doing but they had built this robotic control system like android app so they built an android app so that they can control a robot that they were like an industrial robot like big one in Germany and they were like playing with it with their with their phone and I was like that's yes that but if he would have sent that to recruiter they would have totally rejected it because it wasn't whatever framework version something something so definitely talk with the hiring manager as well is there anyone here who's a recruiter no seriously don't be afraid because okay then so yeah go ahead raise your hand I'm going to share with you in let me try to squeeze it in two minutes how recruitment works okay so when company is hire recruiters different recruiters have different tiers there's a tier one recruiter tier two recruiter tier three recruiter so the tier one recruiter gets the job right immediately when it's available and the tier two recruiter gets it a week after the tier three recruiter gets it a week after so a tier one recruiter has an advantage right so how do you keep as a tier one a recruiter you give them good resumes every recruiter can only give maybe five resumes so you can't spam them resumes but you can only give five so from the recruiter's perspective how do they know if these resumes make the mark to them because they're not technical they can't they don't really have too much in them experience for them it's all about meeting the checkmarks so that's how a recruiter works so therefore going through the recruiter route is actually a lower way of success and you know as people say going through the referral route if you know friends and technology companies talk to them try to get them give them your resume because they know what they want as a teammate or they know who they want to hire you're going to have a better chance of success so let me share with you an example when I'm trying to make a career change I sent out okay the interview hit rate was 10% so for every 10 resumes I sent out I only had one interview invite when I went through the referral route the interview hit rate was 100% so it's like yeah I can't say what he did actually he looked at it and he started to search if someone actually had a similar background to him and then he'll now get in the job with like a little scientist so now you're going to send out an invitation to try to see how one can have a catch up with these guys of course and you'll probably send out 10 emails when no one come back if you send out the 50 because one or two emails come back then you probably want to catch up like that's like sharing with him no understanding one side you probably learn from their diet the other side you probably get opportunity he started referring to some junior positions if possible because today I'm working for a lot of MNC's company recruiter is one side and many times their priority will be someone who refer this guy to me so I know this guy we have an internal referring program running everything so if I'm working for this company I'm going to an MNC I know this guy his potential is good your chance of getting to be almost 100% yeah actually to add on to that right I would say almost all the jobs that I have till now are all through referrals yeah I can simply say that it's really 90% through referrals I think that's the reason that's one of the reason why it's good to set up a community and and ready go and go out and network to other people talk to other people don't show other people what you what you have done because this is a technical community everyone will have sort of an idea as to what is required for for that particular job and they will be the the right people to look at you and say that yeah I think you have the suitable skills for our for our company I think that's that's that's one of the aim of having having such a such a community to really build up your network yeah I think a lazara more than 90% of our hires are referred for the data science team okay so I want to touch on two points that Koo and Adam made one point they made is I think who made this it's hard to be a data scientist if you don't have the basic software did I rephrase that correctly software experience software or I think who or Adam yeah Adam said that it's hard to be a data scientist one of the basic software okay so software engineering is one part of it and Adam also said documented righted so the essentially written communication so what is the one thing that you think the audience can do to improve software engineering maybe Adam one thing that they can do for an hour deal so there are a lot of sites out there where you can find programming challenges that are relatively small relatively short so one thing that a lot of people have trouble with is they don't know where to start so they don't have any ideas so that they they say okay well I need to work on my my software development skills I need to be better at that but what kind of project should I do I don't know what to do my advice just go to hacker rank or something and start doing the practice exercises they have tons of them all in all different areas they have they even have a whole section I think on on machine learning and then you can use all kinds of different languages Python Java C++ go Haskell Scala Clojure whatever you want so that's kind of an easy way to do it it's just really digestible you can try to to work on one of those or two of those a day or whatever you want so that's one thing and on the writing the the best way to get better at writing is to write more and the only reason really that you shouldn't be writing more is because you just physically don't have time to write more it matters a lot I cannot overstate how broad the effect is of having content and things that you've written out there especially if you're thinking of transitioning into a data scientist role the reason being that let's say that you do the networking the writing and the web presence is going to help you let's say that you get the interview that's going to help you because people are going to look you up and they're going to form interview questions based on stuff you've already done which is going to make your interview process a lot smoother and more comfortable and then let's say that it works out and they want to hire you well now they can't use that nice trick where they say well you're transitioning and we're not sure about your experience so we're going to have to start you at a lower salary like no I've already done all this stuff before so it gives you so many benefits and if if you're writing about things that you're doing anyway the investment for you in terms of time is kind of minimal for the payoff that you get so I 100% would say write more every everything you do every project you do I did I was a couple of weeks ago I wrote like a Twitter bot which I haven't done in a couple years so I want to do another one and I the day or two after I finished it I in the morning I just wrote up everything that I did and put the code on GitHub I just it's part of my process that I do and I can't recommend it enough it gives you so many benefits secondary tertiary just it really it really does cascade what about D8 maybe we can let cool finish you were going to say something yeah actually I just want to sum you up which just start working on the project yeah what project people don't know where to start uh okay for me okay for me my recommendation is this but this is my learning style um I will say the one that really kicked things off for me was actually Andrew Ernst machine learning course which was on youtube was your Singaporean one yeah so I yeah I started from there I I learned along the way as I watched the the videos um then I start because of the keywords that was used to explain the videos like all the calculus stuff the linear algebra stuff uh the cost functions and all these all these sort of provide me the keywords to actually take the next step to do a google search see what comes out from there um so I'll do start a project when you just start something and and move on from there using the keywords and learn to learn from there and now I have actually a long list of projects that I wanted to do one of them I just shared with Adam uh yesterday was actually doing a chessboard uh chessboard yeah so what what one of my retiring plans is to actually create chessboards for all the different chess then I'll like play with it during my retirement years like us are you expecting to retire alone the only only person left on her if you have chess parts you're never alone yeah I have like probably seven or eight ai's to who needs two months when you got ai yeah okay anyway yeah it's about writing and documenting your activities and your experience what about nbe I mean usually for example I'm facing this problem every time when I can't interview they ask me here have you have an account yes I have an account but I'm sorry oh GitHub yeah yeah I have a GitHub account but there is two projects inside and I hope probably 50 of different projects in different companies but they are under nba okay yeah how to afuscate it I don't know how I mean probably you can describe your activities in some absolute way not to disclose I mean I would recommend doing another project on your own time that wouldn't be so it's it's a bit difficult because depending okay so now we get into a little bit employment contract stuff so if you go to work at a company some companies will try to have in their employment contract that any idea you have while you're employed with them is their property I don't know any country in the world where that's actually legally enforceable first of all second of all it sends a horrible message to the employees so that's usually the first thing I try to get stripped out from contracts so if that's your concern that something you work on on your own time will somehow be property of the company there's technically a legal risk there but it's pretty small so if you're talking about it's under nba because somehow the employment contract then that's I wouldn't worry about that and not only that I would probably raise some hell about it too because that's just not cool right I mean it's like saying you can't have your own ideas that doesn't make sense if it's under nba because it was an actual work project that's a different situation like just professionally ethically you'd have to let that go it's you don't even want to get into a situation where you kind of do a modified version of it or something like that and then try to release it because people talk and if someone knows that you're doing modified versions of your work code and then putting it on your github suddenly your your trustworthiness is called in the question which is horrible so then I would in that case I would agree with you you just do your own a bad sign for your employers it would be 100% a bad sign but then we get to a different question which is there's a lot of code in companies that can be open sourced and I've tried to do this in places where I've worked where we actually segment out some of the code that's not proprietary because most codes not it's some kind of business stuff so if it's something like that then you may be able to find some kind of ally internally and start working with them and helping them understand why it would be important and helpful for the company to open source stuff then you can start getting that train rolling usually you have to sell it as a recruiting and retention benefit something that's good for the public image of the company and so on because that's what more of the business people care about but that's another option too or it's like getting the community to help you find bugs and fix them if you open something there they don't have anything to add what platform to publish it do you mean post-it notes VIM so I don't want to start the VIM Emax war but VIM so you mean publishing platform or like or or editor or oh publishing whatever you want anything to get it out there it can be on wordpress you can do github pages you can do static site medium i don't know some other one okay i have a problem with that one because i don't want them owning my content so i want to keep my content so i don't publish on linkedin i don't have a facebook account but i keep all my content so i don't write on linkedin that's just me but the the answer is write anywhere that you will actually write that's that's important so don't pick a place that's going to make it difficult for you to put the content out anywhere you feel comfortable or happy or motivated writing the stuff and publishing it that's the place to choose it's just it's just a tool it's just a channel for you to get your message out sorry yeah so if you if you put the content on wordpress.com or something it will be searchable on google yeah i think so koo mentioned just now that he started off by doing machine learning classes from andro and then he went on to talk to us in algebra etc etc so there are two questions here that are quite related to this so why should we learn the math or computer science behind the tools slash models we use after all most of them off the shelf seem to work well enough we just need to optimize the parameters generally for true cost validation and deploy it and then the other question is what is the relationship between the theory of machine learning and the practice of machine learning for deployment from the point of an engineer so someone who designs and builds stuff and deploys it as opposed to an academic doing research so maybe the first question first why should we learn the math and the computer science the fundamentals like why not about old notation or linear algebra i think i think having a good understanding of max right allows you or gives you a lot of flexibility in terms of like treating your algorithms later on when you actually have a different situations so what i mean by different situation is because let's face it i mean even even company in the same industry right will have different business models different business policies regulations different kinds of people working inside and all these kind of stuff and you need to be able to understand how to treat all these algorithms so that you'll be useful in a particular context that you are working within so i usually give a very common example is the difference between the rich regression and the lasso regression uh you look at it it's actually the regularized regularization part which is one is absolute the other one is square so what's the difference why why a square what if it's a square right what kind of uh um impact what you have on the regularization and what if it's absolute what kind of what kind of impact would it have so all these will come in the picture and you have to think about how does it how all these mathematics will then translate into business uh into the business context what what kind of impact we have on what kind of impact we have on business so um for any students that i come across in i always recommend them that please don't say that you don't want you don't know max max at all but i have to say if you really want to be very proficient in data science you can't run away from max you can't you really can't run away from from from the max part and i have to say i'm very lucky in a sense that i love max a lot similarly like like i can sit i can sit down and solve a problem and spend two hours on it i don't mind at all i think it's it's it's perfect it's just that maybe there's sometimes a Greek letter sort of like haunts me during my my sleep but but still okay still manageable uh for me and i love it looking looking at all these uh mathematics problems okay so math some people think that the math is the implementation detail right like they think oh well there's just some regularizer and let's see but actually it's the other way around the the algorithm is actually the implementation detail because if you understand the the mathematics that underlies all this stuff then some particular algorithm is just some application so if you get the linear algebra and the calculus and all these things then a lot of machine learning things for me as someone who studied math are kind of like more of an implementation detail and that's very powerful because you can transition between a lot of different concepts very easily whereas if you don't understand the underlying mathematics and you're learning everything case by case then sometimes it can become very difficult that's one thing but the main problem that people face when they're learning math stuff is that it's hard and for some reason they don't think it should be hard and so then they think they're not good at math because it's hard and i can't speak for everyone but for me math is really hard i'm i'm not the best at math i i did uh an undergrad and a master's in applied math and i spent a lot of time around a lot of people who are a lot better at math than i was and the thing that i learned is one of the things i learned is that if you are doing math and you feel stupid that's okay that's okay there is no problem with that there's no problem with it it's it's an area where when you're learning you have to struggle with it it's much different than uh some other areas where you can read a book and just kind of lightly skim the book maybe you're only halfway paying attention and you sort of get math is not like that you have to focus a lot and you have to mentally struggle a lot but once you do that the understanding you get is really profound so what i see most often as the impediment to people learning math is the fact that they don't understand that it's going to be hard and that doesn't make them bad at math it doesn't make them incapable it doesn't make them stupid it's hard for me so i do it anyway and maybe that comes back to the discipline thing like you said sit for two hours there have been times where i sat for two hours with the problem and i still didn't fix it i still didn't solve it and i came back the next day and i i tried to do it again and then eventually i figured it out and for me math is like that you you have to struggle with it but once you do that the understanding that you can get for these machine learning applications is really profound it's really okay so i think we have time for one last question on the general data science track i guess this question comes out a lot masters in data analytics causes cause the great deal recently there are certifications from uh microsoft like the microsoft professional data science program coserra data science tracks they are more affordable how useful is how useful is it and is it worth it both time and money assuming one finishes the course for your own development and for employably okay i think it's a trap question for me actually you're on camera don't forget yeah i know yeah i i will say i'll say it feels a it feels a different niche uh i mean there are some people who i want a master's degree and i also want to learn data science at the same time um then then of course a a degree program uh helps uh disregarding the cost um so having said that uh i will say it feels a different niche it's up to you i mean for for me i i kind of think i have self-discipline um to actually sit down and look at the problem call it out uh and all this and so for me maybe not a master's degree but anyway i have my i have my masters really which is in uh in business admin yeah so i will say it feels a different niche so you have to ask yourself the question of uh a balance between cost and value i mean as a economist background i i will say i think the general question is like these online certifications data science tracks from costera all this um is it worth the time and effort do you really get value of it for your own learning and for employability and also there's a fact that i think if you throw a rock out in singapore almost everyone has one all right so many people have doing it because idea is given out of subsidy in case you don't know about it now you know if you finish the data science track you get five hundred dollars back which is the cost of it and you get five hundred dollars more so you earn money by doing courses but the program ended yet not the program ended yeah program ended okay sorry guys yeah so i mean adis what's your take on this like for people out here who are considering so learning is always a good idea and it's never wasted that's the first thing learning is never wasted whether you want to pay for the learning or not is kind of your call that that's really what it is the primary benefit that i see from these kind of programs is that it allows people a bit of structure so that they can actually do some of the projects and some of the writing and stuff that we talked about earlier some people have a hard time doing that without a bit more structure and that's fine i see the same benefit to some of the data science kind of boot camps or programs is you you get a distilled curriculum that's focused and you do projects as part of the program and those projects get put up on the web so that employers and the community can see them so you get that kind of exposure and practice and usually there's some kind of presentation component as well and that matters so there are positive effects from from doing something like that but like kusse you don't necessarily have to go through that program to do it it may be better for you from a time or a money perspective to do those things on your own i think there's a there's a github repository somewhere i can't remember exactly where but i think it's called the open source data science masters or something like yes and it's got all kinds of stuff in there a lot of these a lot of these programs make their curriculum freely available the same thing for a lot of the boot camps so like data science retreat in berlin or i think zipfian academy insight all of these places they they make it available so you can actually see what they're doing so then you have to ask the question well what's the benefit of paying for it and the benefit of paying for it is basically the structure and the network i would say you get a little more guidance you get more structure to the learning part and also to kind of the networking and the interaction with the community so if you can do those things on your own then you don't really need you don't really need the program if you're talking about employability it's the same thing so most i can't speak for everyone but for me if i'm in a situation where i'm a hiring manager the fact that someone has a certification or a degree or whatever is it's an indicator or a signal but it's it's an it's not a sufficient condition of employability and i mean that in the in the mathematical sense i'm not going to employ someone just because they have a certificate or a degree that might give interesting topics to explore or talk about during the interview but i'm more interested in their projects i'm more interested in the other kind of work that they've done and the things they've finished and how they communicate with other people things like that so i wouldn't i wouldn't advise anyone and say yes you should definitely get one of these certificates but if it's going to help you and give you the structure you need to put together that public portfolio of projects and stuff then you have to make the call whether or not it's worth it for you okay so i think we have four or five more questions that i related to general data science i think okay so here's a challenge in conciseness and succinctness for our panelists i'm going to ask you the question try to answer it within 30 seconds okay with what are the three must-know things that you expect from data scientists on your team cool i'll take 15 seconds to think uh you can you can take you only have 30 seconds to answer you can think as long as you want okay let's give cool a break Adam okay curiosity and tenacity that's it i only need to so so tenacity tenacity slash i'll say interpersonal skills uh last also logic thinking it's amazing none of them said python spa okay so what is the next big thing in machine learning slash data science how quickly a lot of it's going to evaporate i'll say ai actually i mean i'm not going to go with the it's not going with the trends and all this but i think ai is something that we we will move towards next uh when it comes to um data science but having said that i'm not saying that they're equal i think i've put up an article before or on the facebook page is these three are very separate you have 10 more seconds uh so read out that article yeah is it for cool's article or where uh no it's not mine it's actually from from data science center if i'm okay yeah so so it actually separates all these three out please read them out because nowadays i'm seeing a lot of people saying that they know ai but actually what they know is machine learning yeah so i'll i'll give my real answer now my my real answer is i've been a big fan of streaming analytics and reinforcement learning for a long time and i think that's going to become a more important issue as people want to analyze data while it's in motion because it just becomes too expensive to analyze it after it's been stored so if you want my guess uh also a lot of people talk about internet things that's kind of part of it that that's not really the main issue but reinforcement learning in general and and streaming and learning okay it's kind of related to the previous adept's first response this data science just a short term hype is this also a 30 second one because that's that's a that's a more than a 30 second 30 second short term hype in its current form yes because it's going to change what's its current form so you get a lot of people who are selling the result instead of the process so people are selling this concept of data science because they want to have what companies who talk about data science have they want to have what google has they want to have what facebook has they're not actually talking about the process of getting there they just think they're gonna whatever they're gonna set up a hadoop cluster like google and then they're gonna make money like google and companies are figuring out that it doesn't work that way so i think in that sense expectations are going to be mitigated and more appropriate which is good for all of us actually so is it short term hype there is some hype right now i think it will die down but it's not going to go away cool uh i agree with adam on the last sentence so but generally when i share with other people i said that um there are three questions you like you need to ask yourself will technology improve answer is yes technology will improve second thing is with technology improve right do you think companies are going to stop collecting data i don't think so that's that's an i yes because companies are going to still keep on collecting data so the third one is i think going to let the data sit down there and rot but if it rocks um until it's no again so i'll say the science is here to stay but yes it could be a hype for now but it's definitely here to stay so i'm overwhelmed by how fast the industry moves how do i keep up what should i prioritize on anyone whatever that keeps your job so i'll disagree i'll say i'll say the thing that's going to keep your job is important but it's not the the main focus the main focus i always advise people is to to look at transferable skills that's going to be communication of math because that will take you to any other job so if you if you understand the cs fundamentals and the math fundamentals with new languages or frameworks or algorithms or whatever come out you can read the research paper and you'll get it pretty quickly so that's going to be a lot better time investment and because who cares if there's some new javascript framework for front development i mean that might be useful or interesting but that's not something that's going to make me really concerned so the main thing is to focus on the fundamental transferable skills because then you can learn these new things very very quickly okay so another question how do you see the market for data science slash data scientists i think you answered that already how does it compare from three years ago and how do you anticipate three years from now it's going to grow the last question which sectors in particular are receptive or have the greatest potential sorry which one which sectors have the greatest potential for data science i think if you if you're looking for a management perspective like profit centers are the one that will actually definitely have a lot of potential when it comes to data science but having said that there's a there is an outlier and the outlier is actually recent compliance which is seen as a core center but recent compliance are um it's a place where i think uh data science would flourish um and you can safely say that so someone was telling me this actually if you if you were to choose like will you work in a profit center will you work in a core center i said okay if you want to work in a core center please make sure it's recent compliance even during a recession right there's no way to go cut down the compliance department so you can choose so make sure that you go to a profit center that's where there'll be a lot more data science uh oh okay then yeah if it's a core center make sure it's compliance and risk done so can you restate the question which sectors in particular are receptive or have potential for data science in the next three or five years okay so this is kind of well i don't know if it's an unpopular perspective but insurance actuarial studies is kind of like the original data science in my opinion it's like application of statistics and math to a particular business problem it's been around forever and insurance companies have a lot of money and they know that they're sort of floundering the other thing is a lot of new startups are coming up and building completely new insurance products that didn't exist five years ago because we have data now so we can calculate risk on things that we weren't able to calculate before so insurance is an interesting one fintech's getting a lot of press right now i'm less optimistic on that one i i think a lot of it's going to be algorithmic trading and things but i think it's still too early to say that that's going to come but insurance definitely i see i see that's one and i would agree with with coup that if you're going to go into a company go for a profit center cost center no chance no chance because companies respond a lot better and people respond a lot better to areas where you can make them more money than when you can save the money you can only save a finite amount of money but you can increase revenue by a much different amount okay i know i know a lot of people are looking forward to the dealership part two more questions with the growing competition a few how do i stand out from the crowd cool you can't just say get a degree from where yeah how do you stand out for a crowd i would say if you're hiring what makes someone stand out i will i will actually i will look at the projects i will look at the projects that they do and from there i would see how interesting or how complex the project is i mean of course the more complex the more complexity that is involved then i'll definitely probe further and to actually sort of like support that this person is actually going to be outstanding as compared to the others yeah so i agree it's about the projects i think the the other issue to consider and how you want to take advantage of this in your own situation is up to you but i do see more fragmentation in the field than i did a few years ago so three five years ago people would just hire a data scientist and they would do kind of everything and now depending on the company you might have someone who's not a data scientist but they're a recommender system was engineer or something much more specific much more focused so if you have areas like that which are interesting for you it could make sense to spend more time focusing on those but other than that it really it really is the projects standing out well i mean you're already standing out because you're here and everybody else is maybe still sleeping so the fact that you're doing stuff while other people are sleeping is a is a good approach if you keep doing that then everything else kind of take care of itself okay so this is a tough question to answer in 30 seconds but how should one prepare for a data science interview maybe there's no way to answer this two projects projects showcase your project showcase your project yes i think the projects will be the one that really tells your employer what you can do what you can't do for now it gives you a very good gauge as to your current skill set and at least the the hiring managers will be able to ask the right question and also be able to gauge whether you are going to be a good fit into the company as well yeah so two things i would say one practice and learn how to guide conversations once you get into an interview you should take the first opportunity to turn the interview around so that you start interviewing them and that's a skill and you can learn it and and it's something that is extremely valuable and once you have the ability to steer conversations like that you will have a lot more success in interviews because you can control which questions get asked or which questions you answer or how long you spend on different questions so that's one thing the other thing is if you want to be successful in an interview it helps if the interview is just a formality so if you already kind of know the company and the people and what they're working on and the projects and the challenges and problems and all that stuff that gives you a lot of opportunity to guide the interview and you can focus on how you can contribute to those things so one thing is just conversational skills like learning how to guide a conversation and you can do that at an interview you can run the interview even if you're the one applying for the job that's totally fine there's nothing wrong with that then the last thing as a side note is take the mental perspective that you are the one who's interviewing the company and they have to convince you they need to sell you on why you should be there even even if you don't think that's true just take that mental perspective and it will really change the way you conduct your your interviews I think Adam makes a good point so a lot of companies claim they want to hire a data scientist and then they join if they realize they've been working on tabletash boards for a year so I guess most people here or maybe some people here don't want to do that so as an interviewee as an applicant you really have to ask them how's your database like what will I be doing in the first three months what do you want me to deliver what do you want me to achieve and if the achievement is management visibility of metrics then wrong yeah I'll just say thank you and walk off the top well we don't literally okay so that's enough time for general data science questions of course our will you guys be around after this our panels will still be around after this so now I'm going to move on to the more leadership related questions so one question is how do I deal with non-technical superiors or leaders who are not technical but think they are and try to overly manage things so I'm sure no one none of us here have a has a has a boss like that right so I think I think this this question comes out like quite a few times when I was discussing with with a few friends so generally I think we need to find out a bit more as to why they behave that way by coffee by lunch yeah we have to find out a bit more why why they behave that way I would say then from there then we can actually really find a good solutions to it so I have what I would have to say it really depends on the on the person and on the on the situation but more importantly is I feel if a person actually behave that way right don't openly say that he behaved that way that's the most important don't openly do that then go a bit separate from the rest of the people but and if the boss is actually something that's receptive then slowly bring across the message don't make a heavy impact but slowly sip in your message yeah that's the most important because once you get them more defensive right there's no way you go get through the tix car anymore yes that is true okay so I would say first take a very different mindset about dealing with them and feel sorry for them which sounds kind of weird but stay with me for a second so you're not the only one who knows they're doing this so other people know this too and imagine they may be aren't aware of this but imagine what it's like to have a bunch of people thinking that you're just this horrible person that doesn't know anything you should feel bad for them right it really sucks it really sucks so the first thing I would say is mindset shift start feeling bad for them then start figuring out how you can try to help which should not be such an issue now that you've had this mindset shift right because now you feel bad for them so you know you're a good person so you want to help them out that's the way I would approach it then you can start doing things that might be useful for them and try to over deliver because the reason they're probably micromanaging is because they don't trust why don't they trust because maybe they have trust issues who knows but the point is you got to build that trust and that means you have to find ways to make them look good remember you're not here to be right you're here to win and if that means you make your boss look good for a while so that they get promoted or go somewhere else or give you more freedom or you have room to operate you win so that mindset shift helps to suppress your ego so that you're not constantly trying to be right or you know defensive or whatever with the boss you feel bad for them because it's a bad situation for them you do what you can to help them which then helps to build up the trust and then you start to win because you get the autonomy and the freedom so step one mindset change I think it's one way to prevent this problem is also like we mentioned the interview interview your potential boss if he's going to be an idiot who doesn't know anything but just says hey the best way to increase conversion is doing something ridiculous so the interview is really the best time for you to figure out if your teammates are good if your boss is good and you want to work with them and I think to summarize what Adam and Koo said take ownership it's not their fault that they are micromanaging you it's your fault that you're not communicating to them well enough in a way that they can understand and you're not persuading them or what's the right course of action okay so next question how would someone know if they are more suitable for the technical track versus the management track other than trying it so I mean of course anyone can just try to be a manager and then oh I suck at this I prefer to just write code not attend meetings or someone can just oh I don't have to write code is there any way that you yourself feel that hey this is what I like I would say not don't try it I think regardless you really have to you have to try it but I think you have to find an environment where the cost of trying it is lower I think that would that would be something that makes more sense so where are these environments that that that allows you to actually try out at a lower cost and I would say doing group projects during your during your lessons or doing in I mean for the students for the students yeah but for those who are mid career now maybe gather a group of people to work on a similar projects then talk to these people interact with these people network with these people then I think it will be good so it helps with the network it helps with your your reputation and yeah it could be a lower cost I'm not saying that it won't be so try to be a leader but not at work yeah not at work maybe somewhere else where it's not like a 24-7 thing that you need to be yeah so I would say first you don't necessarily have to choose that's that's one thing there are some companies where you don't have to choose so it depends on where you work hopefully you work at a company where they value their technical people so there's a technical career progression that doesn't require you to go do management stuff the other thing I would say as some kind of red flags for yourself internally if you're thinking that you want to do leadership and management stuff because you want more money or better title or something like this please don't you're going to make everything worse for all of us it's not about that the number one thing is you have to do it because you want to help the people and you really have to care about them like you really have to care you have to they have to be like your own children it's like that and if if you're not really that interested to deal with people then don't do it it's it's pretty straightforward it's not it's not a big deal it's not a problem I've done experiments before where people have come to me and said I'm kind of interested to do some leadership stuff but I don't know if I'll be any good at it and I always frame this as an experiment I say okay that's no big deal we can try this out for like two three months we'll see how it goes and if you like it we don't have to keep doing it and if you like it and it works out then we can keep it up so frame it as an experiment and I do this with a lot of things people are really hesitant to do things if they think it might fail but if you call it an experiment explicitly then for whatever reason suddenly people are okay with it so frame everything is an experiment it will really cut down on the defensiveness and then you can like who said you can try it out you can try it out because it's an experiment right why not the other thing is like we were talking about earlier leadership is influence so you don't have to be like promoted or put into some kind of management role to actually do leadership stuff you know you start by leading yourself every day and the influence that you have with your bosses with your peers whatever that's also an opportunity for you to exercise leadership and you can kind of figure out if that's something you want to invest in or you don't and either way is fine but what I see as a problem for a lot of people is when they do it for money or kind of career fame things like this I've had people literally come to me and say oh all my friends are bugging me because I'm not a senior architect something you can you can cut my salary just give me a better title and make me a manager it's like no no so it does happen so if that's the reasoning maybe not the best idea but if you want to help people and take care of people then it's a it's a pretty nice pretty nice gig yeah based on Alan's response I'm actually quite surprised that care for people is actually not on your slide 11 humility and discipline if you're if you're humble then you're able to do that's true at google they found that the number one trade that predicts whether a manager would be a good or not in terms of team performance when a team likes them it's actually whether a manager actually cares for them whether the team actually feels that the manager cares for them they ask them daily hey what blockers you have or if something happened in the family they're like oh you know what yeah so I think that's the main thing I think I was thinking of is one of the other one of the skill I was I was one of the characteristics right that I really think would help a data scientist is actually empathy being able to put yourself in the other person shoe so why I said that is I said for instance if you're doing a presentation you're able to put yourself in the management shoes able to think like what do they really want from the presentation that helps if your team lead if you have empathy you're able to put yourself in the other person shoe and anytime when you give them like instructions or you give them any like constraints and all this at least you'll get a buy-in and empathy when you talk to other stakeholders like many people in the data engineering department you're able to communicate to them you're empathize with the constraints that they're working with they are more willing to actually work with you I would say so one of the key people skill that I feel the other scientist should have is actually empathy being able to put yourself in the other person shoe okay so in line with this what is the sacrifice required when you move from a technical position to a management position so I can think off the top of my head oh I need to attend more meetings I won't be able to write as much code I won't build things myself as much I have to sell I have to sell my team so I can get more budget I have to sell my team so I can get more money to buy clusters some people don't like to sell stuff we include it so but what do you think so there can be a lot of sacrifices but if you enjoy what you're doing then it's not really a sacrifice so it's more like a change so I wouldn't be opposed to having a job where I could write code all day and just kind of chill out with my editor that'd be nice but what I found is that eventually the kind of impacts I can make in a company are limited for me as an individual because I cannot scale my brain any more than the code that I can write so leading teams is kind of a way to work together with other people and scale my brain out and teams of teams and so on and so on so if you make that choice yeah you're making a trade-off you're going to have to maybe go to more meetings you're going to have to deal with more people more politics sometimes as the sacrifices can be severe and political there there have been cases where I've had to use a lot of political capital to give people raises because it was the right thing to do as a leader but it cost me a lot personally there have been cases where I have left jobs over the years so that other people wouldn't have to take pay cuts and those are the kind of things you have to be willing to do you literally have to be willing to put other people before yourself and if you're not willing to do that then you may want to ask whether or not you're ready for a leadership in a management role because it goes deep when you care about the people if you're in a situation where you're leading a big team of people and you know that there has to be a budget cut and it's going to be your salary or someone else you're going to have to make the call and you're going to have to live with whatever that decision is so sometimes there are real sacrifices like that but in general it's extremely rewarding so if you want to take care of people like that do it you you won't regret it. What about you? I kind of agree with what Adam has mentioned so far so I don't think there's any sacrifice if it's something that you enjoy doing and if you want to you can see yourself as a mentoring role as well as a leader. First I would see myself as a mentor when I if I'm in a leadership position I would love to grow many of minions. Many people of me decline so that's how we can actually scale up faster and faster. Better or not that depends but definitely faster so I don't see that as a sacrifice if it's something that you are you love to do and I would say most of the other scientists like to create impacts positive impacts especially so I think being put in a leadership leadership position actually gives you a position where you can actually create a bigger impact as compared to being a single a single person in a team. Okay so with that in mind what is the different skill sets required for someone transitioning to a leadership or management position as opposed to being a data scientist so being a data scientist you need software communication I assume leadership also requires a lot of communication if not more but what else is different so if people want to move into that kind of role how should they prepare for it so they don't fail. Okay if you're going to get into a leadership position right you definitely have to communicate to more people more stakeholders as compared to individuals in your team and having said that right I think it's always good again I said you need to be able to empathize with your different stakeholders but empathy is such a general word right how to even go about doing I feel is you need to be able to understand the technical skills that are involved with the different stakeholders like if let's say if you're a data engineer what are the what are the kind of technical skills that's important what kind of constraints do they have and all these so as a leader of a data scientist as a leader of a team of data scientists I would say you need to even know more than what your your current team knows but of course not when I say no more doesn't mean that you have to know more in that but more of like you have to be a bit more generalist rather than a a specialist ready so I think the communication becomes more important and not just written communication but also interacting directly with people so as an example I worked with somebody I don't know five years ago or something and this person wanted to transition into a leadership role but they had a very difficult time communicating verbally they they were very quiet person so even if you were in a meeting with them at a table a normal size table it would become difficult to actually understand what they were saying and it seems like that's a small thing but it actually matters a lot when it comes to humans not just from a communication kind of efficiency perspective but also from a psychological perspective if if you're sitting with someone and they're quiet to the point where you can't even really hear them then it can be difficult for a lot of people to take them seriously as a leader so we had to come up with exercises so I would take him into a really big room and I would stand on the total other end of the room and we would have a conversation like this and I helped him with some breathing exercises that he could do at home to become more comfortable taking larger breaths and things like this and there are a bunch of tiny things but these kind of communication things really add up written verbal you have to be able to project your voice and give presentations which is something that a lot of people who are in a development team maybe aren't used to doing and aren't comfortable doing you have to like who said you have to have a different perspective a broader perspective you have to understand what the business wants and why which is maybe different you have to understand that a lot of times the business doesn't care about anything technical it's totally irrelevant to them if you start talking about technical stuff you're basically wasting their time and making them frustrated and they don't feel smart and then you've already lost the audience it's already bad at that point everything's done so the way that you frame the communication becomes a lot more important tact becomes a lot more important you're you're interacting with people who have a lot more authority both real and imagined so that's that's difficult so I would say those are all things that you need to change how does one practice that practice what gain the ability to communicate interact with people who are not easy to communicate interact with how does one gain tact this political correctness so tact is like empathy I mean if you're empathetic then you have a bit more tact and the other thing I'll go back to it again it's like you're not there to be right you're there to win right and those are not the same and many people don't care that they want to feel good and they want to feel right even if they don't want to necessarily even be right but they just want to feel right and so finding ways empathetically where you can do that and still make progress that's how you win and it takes practice basically I mean there's no there's no magic to it you just have to interact with people more that's that happened to me years and years ago I realized okay technically I'm really comfortable but what's actually holding me back and what's an impediment for me is my interpersonal skills I need to get more talks I need to talk with more people I need to understand more about people because if you're a tech person the humans are like undocumented api you're just sort of like hope and you try to figure out and sometimes you hit this endpoint and you get this response and sometimes not and you don't really understand why it's not really really reliable yeah it's not necessarily reliable and it's like that's basically how I approach the problem is like every time I'm interacting with a person I'm basically exploring an undocumented api like I get the human api but there are sometimes the endpoints aren't quite what you expect and so that's kind of how I approach it and it really is case by case and Alex earlier was saying have one-to-ones with your people it's one of the best tools that you have the leader and it's true and the reason it's true is because people are different there are some things foundationally that are the same but motivators are different reactions to certain things are different some people are great under stress some people aren't some people you want to present information to a certain way and other people would become totally defensive and wouldn't be able to accept it at all so it's really just for me it was changing the way I perceive human interaction instead of exchanging information because that's how I used to perceive it I was like okay this is an information exchange you know and I realized we're doing UDP not TCP so it's totally different right you really have to just change your perception kind of like with the boss story right if you start to feel bad for the boss then it changes the way you interact with with the bad boss right but it's the same with people so I think this question may not arise so much when a data scientist is trying to transition into a leadership role because they already have the basic technical skill sets right but for companies that are trying to hire a data science head head of data science this technical skill set is important so let me let me let me phrase it this way on one hand many people say that yes you hired a head of data science and he's supposed to hire people who are better than right because that's you hire people better than you so that they can do the job better on the other hand at google when they hire project managers or program managers there's a requirement that the project or program manager has to be at least as technical as the team I don't know what that means as technical as the median of the team or the least technical person of the team but what what's your take on that okay I think that's a good leader the first thing you want to do is to sort of like secure the secure the trust of your team I think that's the most important if you don't even secure the trust of your team like I think your your leadership position won't be sustainable so having said that right so what's important to build that trust I think having a good understanding of technical skills have that because then again you'll be able to again goes back to being uh empathized with uh the the required amount of work I mean how many of us have come across this situation where wow my managers just my managers just keep one telling me to do this a b c d e f g and all this while all these requirements but have you done it before it's like you know what you you understand what I mean it's like if your manager if you feel that your manager doesn't understand what you're doing you don't buy in to your manager you don't trust your manager anymore so for me I feel that the technical skills is a requirement actually for for at least for for leader of a data science team so that you can actually be able to understand the work required and also get the buy-in that's one second thing is also easier for you to to understand the requirements on the team and be able to request for the necessary resource outside the team so how does that gel so for example cool is very good at risk management and compliance right and he works in the company so suddenly now he's head of data science and he needs to find someone to help him with data science in a marketing problem so cool definitely doesn't have the skills as for that he definitely is not as technical as the person so will the person still be able to trust cool okay if the trust is not that I think the second thing will be you need to show other people that you're receptive to ideas I feel that if a manager doesn't show that you're receptive to ideas right the floor ideas to you will actually slow down and in fact maybe just stop totally and what will happen is that there will be a lot of back talking about you or this leader sucks and all this and and the morale of the the team actually will drop will drop after that so so if you're ready to really point out I think it's a matter of you have to be give as a leader you have to give a perception that you are receptive to ideas yeah it's just like who said it comes down to technical empathy if you don't understand the burden on the team then it's really difficult to manage both the expectations of the team and the expectations of the rest of the business yeah so I think some level of technical capability is required with that being said though I I don't think it's extremely high so as a leader I I in fact would say I might be using my time poorly if I think I need to be as capable and as much of an expert as every person in my team in every area especially as you get into larger teams so if you have a team of 50 100 200 people they're going to be there's a lot more expertise in that team than you have so I think being comfortable like who said with that fact and saying I'm not the expert I have an idea I can try to help and provide guidance but it's up to you as a team to figure out the specifics and I want to help you do that that's a much better approach so to do that you don't need so much technical background but definitely enough so that you understand the struggles that people have with their work and that's why when I'm when I'm advising companies and I talk with CTOs for example they say oh I'm so busy I don't have any time to write code I say well you need to carve out 20 percent or so of your time so that you can actually be in the code so that you can understand what people are dealing with so that when they complain about technical debt or some test coverage being poor or whatever it is that you actually have some empathy for them instead of that just being a metric so I think it's it's important but it's not critical to be as technical as the medium technical person on the team or something so there's a minimum threshold so you can understand the constraints your team faces and you can clear blockers for them as well as you cannot explain and not oversell through your stakeholders as well okay so then one question is about more managing a team so if I have if my company is set up on a bell curve and I think every one of my team is a superstar and some people are going to get gigantic razors some people are going to get three percent some people are not going to get razors I don't think that's fair what can I do want a break I think yeah okay cool so this is a common thing different companies have different performance review processes and are we still on 32nd clock no okay but we will stop at 12 30 okay at 12 30 okay so this and there's one more question up there's a little there's a little more to this one because it goes into salaries and it's important for people to kind of understand this a bit so I wrote an article a while ago I think it was called like uh what's your current salary or something like this because HR departments and recruiters love to ask this question oh what salary are y'all know what's your current salary never ever tell them here's the reason when you have this kind of stack ranking situation or bell curve or whatever it is the companies pay money to go out into the market and figure out what the salary distribution looks like for your job so there are companies out there who the only product they sell is basically the information on salary distributions so the company will map your job onto this kind of template job and then this company will say okay well for this template job the 50th percentile salary is whatever and they'll give quartiles or whatever they give and then the company will decide based on that what kind of raises they want to do and they'll say we want to get our people to p 45 which is like 45th percentile is sort of where we're going to sit and companies will never disclose this because if they did then everyone would know that they're underpaid for example so the issue with the ranking actually goes one level deeper and it's about how HR departments and companies set their performance review process and the salary adjustments that come from that then you get to your question of if you have this process and you don't think it's fair what can you do about it as a leader and as an employee so depends on the size of the company if you work at a massive company that does this just leave really there's you you're not going to be able to make much progress there and as a side note for people who are in like the first 10 years or so of their career don't worry about getting a raise at your job change your job seriously don't stay at your job you're going to get a much bigger raise and a much bigger increase in responsibility by changing the company you could get 30 40 50 or more percent raise by changing the job if you stay at a company they're benchmarked to that distribution they're probably not going to give you more than a few percent five six maybe depending i mentioned there was a time where i had to expend a lot of political capital to get raises for a whole team almost there were people who got 30% raises we did a market adjustment for people and it was needed but you have to have a leader who's really willing to do battle for you to make that happen so if you're at a big company just leave and if you're at a if you're in the first years of your career also leave every two years ish three years ish you're not going to learn much more after that at that particular company and even if you could they're not going to pay you for your experience they're not going to give you market rate for your experience so the thing to consider is what am i worth on the market not what are they paying me what am i worth on the market and if that gap becomes more than about 20 that's when people start moving you can't really change the performance review process as an employee if your company is big enough you can't change it as a leader you can try to play games with it i don't advise that the the issue is that it doesn't appropriately recognize people who have different strengths it puts everyone on the same standard which some people like it because they think oh we're all equal we're not all equal we all have different areas where we can focus so i think there's gonna be a lot of attrition after today no i'm i'm always really open about this and i've i even for people on my team it's like there are limits to what i always go by market rates so for people on my team i always try to make sure that they're set at market rates i don't care if hr says we can only do a four percent raise or whatever it is that's not my problem that's not my problem that's that's a business problem that they did exactly that so let's suppose my company is a velcro system right and fighting with the hr i mean it's a long term battle and a reason of this velcro system causes conflict right because as a data scientist people are smart as an assumption right because people is not smart you wouldn't want to start to work your team but if you create this velcro system and all of us who know the lack of niche quantities of it then how can i in the best shopping interests manage this down my employees because if it's a velcro system i want to put my credit down so that i can get my salary up but that's the way the game is played so as a boss how can i design a system that despite this velcro system my employees will still be happy okay so what else can you do that's that's the question right so what i've done in the past is a lot of companies have these kind of bonus programs that they don't really talk about but if you press hr hard enough as a leader so this is like if you're working directly with hr right then you can say look this raise thing is a bunch crap because my people their market rate is increasing 15 per cent per year and you're telling me how to give them four that's insulting to them and by the way they deserve more than that so is there some kind of performance bonus system that exists and many companies will have a budget set aside just for this purpose and then you can start allocating performance bonus to people on your team even though you can't give them a raise because it's not allowed by the process you can supplement the revenue the income basically for them but you won't be able to change the actual thing but it's a bit frustrating right because it says like a company right if i give somebody a performance i will not pay somebody else what's the doctor's system right it's not no no no no a lot of these performance bonuses it's a fixed budget that's allocated at hr it's whatever five million dollars for performance bonuses right and then it's actually up to different leaders at different levels and they usually have an allocation limit so you can allocate two thousand five thousand whatever dollars and to your people subject to approval of hr so it's not like you take five from one and you give to the other it's not like that usually it's preallocated as part of the budget if the company doesn't but the simple fact is the companies typically don't do market adjustments because the market moves faster than the company does and it doesn't allow them to predict the financial to do financial forecasts and all that stuff so change the job sorry it's it's it's true it's true sorry you can you can fight that's right but especially for for people in our industry the market rates are for data science over tech people are changing at a rate that business people aren't comfortable with because people in sales or marketing their salaries aren't going up like that so then you get to the actual politics of this so when you get to like a board or an exec team level you're going to have people in a finance team they're saying well why are we allocating all this extra budget to tech and it's usually kind of a back and forth so yeah you can you can definitely fight for your team and sometimes you can win but sometimes the best thing you can do for people on your team is to help them find their next better job really and that's again it comes back to the fact are you doing the best thing for the people you care most about the people and of course you have to care about the company first but I tell you if you do the right thing for the person you're automatically doing right thing for the company because if you're not helping them to find that next better job they're going to do it on their own which means they're going to do on their time and they're not going to tell you about it so you have no way to prepare so if that situation is going to transpire anyway why not help them out and be able to have more information about it it's just better for everybody sorry oh of course of course yeah I've never historically that's something that I get asked so when I'm working with companies like how do you manage recruitment and how do you find tech people there are no tech every every company says that there's a tech talent shortage no there's a shortage in the sense that you don't want to pay for the tech talent that's good if you don't want to pay market rates you're not going to get people that's just the way it is so there's no shortage of good people and and so I've never had personally I've never had an issue with that's never been something that was challenging for me so for me I'd be much happier to help that person move on to something that's a better fit for them than I would try and make them stay there and convincing them why it's just it's just not not ideal okay one last question so given that most of the big data hype is still around with senior management whereby okay press a button get an answer increase in revenue how do you as a data scientist bring them from this big data machine learning magic to more realistic expectations so like it's not like we build a cluster today tomorrow revenue goes up to x I think you have to take an education approach I think there's no magic pill for this but I'm pretty sure most management are smart they will they will be able to actually take what you give it to them but again like I said don't put them on defensive don't don't put them on defensive but go from a point of education educating them that there are certain limitations and all this I also understand why they have such a viewpoint as well because sometimes they actually really from somewhere else and they bring to the table and say that oh this is what I've learned but of course they won't say that this is what I've learned they'll just say that this is what I know so so having said that you'll be good to actually understand why they have that kind of thinking what why they will have that kind of thoughts and then move move on from there and start educating them that that this is the kind of expectation you can get from data science talk to them a bit more relate to them a bit more because Jared I heard a lot like for instance there are some management who says that oh data science are you mean 100 accuracy no not 100 accuracy but you get closer but you tell them that you get closer so you try to find like a common point or the same page some a common point and use it as an anchor and then you move move towards each other rather than saying that oh no you're the one you're the one who's not so smart I'm the one who's smarter listen to me that will put them on a defensive rather and somehow it feels most of the time is because of the lack of information because they're not the technical people so buy them lunch it helps I mean if it's if it's like $15 and it makes your job easier once a month why not yeah so I first of all agree with everything at coup said it's really important to interact with those people because again it comes back to influence and who's the one influencing them the salespeople at cladera or you that's that's it the vendors are going to be doing their job they get paid for that they get commissioned for that so if you are not countering that influence with your own perspective an internal domain knowledge of what's actually going on in the company then you're losing you're basically losing the information battle because you don't want to deal with the problem the problem is not going to go away like I said those vendors are going to be around they're good at what they do so take them out to lunch have coffee with them help them understand what can be done and what not and a lot of that can be done before you even work with the company so like what Eugene was saying when you're when you're interviewing really try to understand what they're planning to do and what their expectations are and what their vision is for data science or analytics or whatever it is in their organization and if it's something that seems unreasonable to you then ask try to understand but take that take that approach where you're trying to understand not where you're just saying oh that's not possible because yeah you're just going to end up with a lot of defensiveness so it's a lot of expectation management and the I would say the higher you go in an organizational hierarchy the more time you spend on expectation management the more time you spend because those people are more detached from the realities of the day-to-day technical world inside the organization so it's it's really important to have the influence where you can because it'll make your life a lot more pleasant So it depends on where this comes from so sometimes and it's a simple case of the SAP people deal directly with the CFO and the CFO went directly to somebody else and then they just decided that they were going to start using SAP sometimes it's a compliance issue and that's the way they chose to address the compliance issue for example you have less you have fewer options in that case but if it's a question of convincing people not to use a certain system I typically don't take that approach I try to understand the problem they're trying to solve and if you start asking questions and trying to understand the problem that someone's trying to solve sometimes they reach a point in the discussion where they themselves realize that that tool will not solve the problem which is ideal because then you're not the one telling them they figured out for their own like on their own by themselves which means you don't get the defensiveness that Kuz so one thing that I do quite a lot is I try to reframe things as questions which such that the answers help me understand more why someone reached that conclusion which at the same time helps them realize that actually maybe it wasn't the best conclusion but sometimes because of ego right they refuse to they know that it doesn't work but ego they refuse to oh that's okay if their ego gets in the way that's their problem I mean sometimes the budget has been set aside they argued so much that hey but then they actually can waste of money but they haven't caught me using it because the budget is I mean the you can't control other people's ego right you can only control your reaction to your own ego so if you have your ego in check that's that's basically all you can do and there are cases like Eugene is saying where the budget's been approved or the project's already been going on for however long like the case I mentioned where I went into a company they had an SAP project had been running for five years they couldn't stop it because they'd capitalized all of the development costs which meant that it technically hadn't shown up on the financials as an expense and if they stopped the project then millions of dollars would just hit the financials right like that and this is not unusual in companies and this is why I say and Ku was saying earlier when you start getting at different levels of the organization a lot more matters than the fact that it's SAP or not SAP right if you have 10 million plus dollars that are going to hit your books if you stop a project or you can finish it and never use it and those 10 million dollars don't hit your books all at once does it sound stupid anymore to finish the project and never use it maybe how many people are you gonna have to fire if you have to write a 10 million dollar check it's not nice right so these are the kind of concerns and considerations that start coming up in an executive team for example and then but people in the development team like this is so stupid we just built this and we're not even using it there might be a good reason like the fact that you still have a job so it's really important to understand that kind of stuff and sometimes you can't argue against it or you could but the consequence would just be outrageous right I mean personally if you told me you have to work nine months so that you can ship this even though we're never gonna use it but that means everyone can keep their jobs I'm gonna do it I'm gonna make sure it happens and I'm gonna win and then we're all gonna still have jobs this can be great I'm not gonna argue that oh this was a dumb project I mean five years and so much and so we should stop it now so sometimes you have to just go with it in fact because it's better for everybody okay that's it thank you thank you to our panelists thanks to Eugene oh thank you the moderator yes we don't have to sit down now we can set up