Starting a Career in Data Science with Massimiliano Ungheretti

Hi everyone. My name is Patrick Akio, and if you're interested in how to find your path in today's data science field, this episode is for you. Joining me today is Massimiliano Ngretti. He's a data scientist over at tom-tom, and he's working on Voice Assistant technology, obviously using AI. Enjoy. I asked you this question and

you did listen to a lot of podcasts now. Yeah, although I should admit only a couple of episodes of this one. No, this one is fine. I don't. I don't get offended. Yeah, No, I do. I do. You know what It actually the podcast became part of sort of of a daily routine. Yeah, that really helped me to get a clear break in the stress cycle during the day, middle of the day, 12:00, no matter what. I go outside, OK?

And I listen to something that kind of, you know, breaks the arc of the day and you when you get back, you feel different. And the podcasts or calling family and friends, that's part of that. And you do that standard 12 each day. Yeah, that's really cool. Yeah, I should do that as well. Yeah. Today I moved it up a little bit. Sure. Fair enough. Thank you. Yeah, that's really nice. And have you have you done that? For how long?

I think it's it's during during the pandemic, it became much more important to me. Before that, I would occasionally do it, sort of only when I was too stressed and I realized, you know what, it's better to use it in maintenance mode. Yeah. Yeah. But before that, I was trying to be very social during lunch in the office setting. Yeah. So it there wasn't a room room in my life. Yeah. And now there is. Yeah.

That makes sense. I do the social bit, and especially when working from home, like my partner's also there. And we would either cook and have lunch together or already have something left over, warm it up and still have lunch together and then do more of the social dynamic and then get back to work. Yeah, but when you're at the office, you're, I feel like I'm always on. Yeah. It's very hard to turn. Yeah, turn off, yeah, yeah.

And then although I think smokers do get a small. They get smokers break. Yeah, yeah, yeah. When I used to work in operations, I had colleagues that didn't smoke but would take like a how do you say that? A second hand smoke break, Yeah. Well, actually I was at some point in a team where smoking, I started to get the feeling maybe could even get you a little career opportunity because you get more face to FaceTime with the important people that smoked. They're the most social people.

Oof. Yeah, yeah, yeah, it's. But smoking now is kind of out of fashion more and more, don't you think? Although the smokers were the ones that led me into the. They got you in. I I should say something. I was. I was really surprised that you're already sitting there. Did you see me? I had to do double take. I was like. Double tape I saw. Yeah. OK, well, I'm also not wearing glass. Yeah, that was also. Yeah. Yeah, that makes sense.

Well. This reminded me of all of a sudden when I was in my last job at INGI was working in fraud detection, anti money laundering, terrorism financing. So the group that takes security pretty seriously. Yeah. And yeah, I ran into someone who after only a while I found out that they were, you know, someone who tests, you know, who does penetration testing using

social engineering. So they're exactly the people will try to, you know, tailgate or talk themselves into a building through the smokers entrance. Yeah, yeah, get that Interesting talk to me about ING because I

saw on your LinkedIn that that was kind of the first instance where you had the role of a data scientist. Before there I saw you were software engineer. You did your PhD as well. Was ING kind of the first professional experience data science wise? Yeah, I think to a large extent it was. Although before that I was doing a bit of research into machine

learning for physics. That's what got me interested into it. When I had a big, you know, when you have to make the big choice after academia, am I gonna be a professor or am I gonna go join a company? And I remembered that one of the things that excited me most in the last years what I was studying was a project I was doing applying you know read this big machine learning book and apply it to this problem we have in processing all of this data for this particle to Tetris CERN.

Yeah. So that was my first exposure to it doing a project in that and it was clear like man, I gotta do this combine the programming the math all of that data scientist. Yeah. Yeah. So I landed ING because they somehow managed to. They they enticed me with the idea of fetching baddies. At least that's that's what stood out for me. I wasn't super interested in the

hardcore finance stuff. I had already figured that out, although I was interested in it, you know, theoretically or doing it for, you know, reading up for fun. But I couldn't imagine that being my job. So I found a place where I could go catch baddies, but for, you know, in a bank, not for the police or the government. Yeah. Different. Very different environment. Yeah. Yeah. Learned a lot. Yeah. Yeah, Yeah. I think if I look back, I was way, in way over my head. Really. The 1st.

Project starting out? Yes. OK I actually when I came out of uni I applied to ING also for data scientist position. I just didn't get in. I didn't have the work experience. I went through the first round and then the second round, they were like, yeah, we're looking with for someone with like two or three years of experience, I was like, I have zero, yeah. And then I ended up joining operations. So I never actually we did in in university. I did do some data science courses.

I have some Python here and there, not much web development, but a little bit. So data science was always kind of on the path, but just do I think a combination of circumstances. I never actually ventured down a deeper rabbit hole when it came to data science. Are you still interested? I'm still interested actually. And now I joined since 2 weeks at ING, but not as a data scientist now, not even as a software engineer, as a product manager. And now I have data scientists

in the team as well. So I'm going to try and figure out how to work together, because I've never worked with data scientists in the same team. I see. Yeah.

All right. But you said you were, you were in over your head. Was it kind of just to the grandeur of things? Were there too many things going on? Was it because of the environment or why was that? I I had a project where I don't think I got, I really grasped the sheer size of the challenge. OK. So the grand jury, that was the the the very first one, OK. It took a while for that to sort of get more structure, more structure around you.

When you're just starting out, I think a lot of people are not very good with dealing with lots of ambiguity. Add to that some politics maybe, you know, lacking technologies. Even if you have to solve all of it, it's the first thing you do. That's hard. The chances of, you know, gaining momentum of actually, you know, hitting it out of the park every now and then, you know, Yeah, be careful. Yeah, yeah. But stay enthusiastic. So I think my enthusiasm. Got the bet, Got the best of me.

OK, Yeah, Ambition. Probably as well coming out of. Yeah, yeah, yeah. I can see that. Yeah. Yeah, so no, the the there was a senior person who I was supposed to work with, but they left after a couple months. And then of course I could decide to go to a different project or not because it was kind of an internal consultancy. But I decided I wanted to stay on for a bit. Yeah, probably I should have rotated projects and get back to the topic later because it.

Yeah, that took me a while to figure that out. Is that stubbornness? Is that? Yes. And I I yeah. And I asked the question a couple of times to people I respect. OK, how do you figure out when to stop something that you're excited about but you realize it's not happening, say, Well, if you realize it's not happening, you're probably already are ready to stop. But actually knowing when to quit something is the most difficult decision to take at anything.

Starting is easy. Stopping just because there's no momentum. It's also easy because you just got distracted. But when you stop, if you find something exciting but you're maybe not making the progress you want to. Yeah, that's hard. I'm still figuring that out. No, but I don't, I don't think so. There's never going to be an answer that satisfies you, I think, because you're never going to know the other side of that answer.

Exactly. It's counterfexual, as the data scientists call it. Right. Yeah, that's how. Yeah, I didn't know that. It's a whole subfield of data. Science. That's cool. Yeah, yeah, yeah. There's something outside deep learning, huh? Yeah. Yeah, that makes sense that there would be, I mean you you talked about kind of letting go, was that also when you then took the step and moved to TomTom or did you do? No, I just rotated to other

topics. So and actually later on I came back to the topic and did some more small projects in that, Yeah, but that was when the rest of the organization around was maturing a bit towards that. So not, you know, there was actually a place for a meteor in there. Yeah, that makes sense when we talk about, because I'm curious

about comfort, right? Whenever I got into a role, I started out in operations and it took me a while, months, maybe a year down the line before I got actually comfortable in what I was doing, comfortable enough and confident enough to actually get stuff done. And then I moved to software engineering and that that confidence kind of dropped and it dropped probably below 0, 'cause I felt like I was. I was starting behind, I felt like even. But then again, you build up that confidence.

When, when did you feel kind of that confidence when it comes to data science? Was that an ING or when was that? And what project kind of the setting? Well, I think you'll always feel a little bit like an imposter and you'll hear them. The best people in the world talk about that. For me, it goes up and down. Yeah.

So looking a bit for external signals and staying realistic in that, you know, so going like, OK, so if I'm one year in, you know, I'll go ask around what's a normal number of projects? So what is a typical impact that you've had for most people in the first couple of years? Probably. If they really added up, it's not much. Yeah. I also believe it's it's going to be pretty important to place yourself in the situations where the chances are pretty good or if it works out, the impact is

gigantic. Yeah, yeah. And did you have like such an impact where you were like, OK, this is, this is what I want now and in the future? Some of them, yeah. Yeah, for sure. I also had somewhere. I really enjoyed the process. You know, I had this big impact. And then I looked back and I'm like, yeah, I'm really proud of how I did it. Yeah, But I don't think I wanna do this again. OK. Really. Yeah. Why is that?

Because of the type of impact. So this was, for example, a project where moving a certain kind of model by 1% point on whatever metric would make many millions. Really. Yes, because of the sheer size of the problem. So if you, for example, if you're giving out mortgages in, you know an entire country and there's, I don't know how many tons of millions of people and I need mortgages half a million. Yeah, that makes you. Can imagine if you move that by one point.

Yeah, it moves. And I realized that these problems could really grab my attention because it's my mathematical background. But in the end, what I did with it, I realized didn't give me satisfaction the outcome. OK. Yeah. What, What would give you satisfaction then? Because you didn't, probably, Yeah. So then I realized actually like tinkering and building, you get this really nice fuzzy feeling. And then showing it to people.

Yeah, I started to realize that, you know, those those types of things I really love and discovering the cases. So kind of scanning an area of a business and then identifying, look, this thing is really you know, if you frame it this way, you know it becomes math and you can optimize it.

Yeah. Is that usually what data science does as well, So kind of that discovery track, because it feels like that goes, that's kind of further in front of what I see usually data scientists focus on, but maybe that's my perception. I think it's also part of the opportunities that the people maybe get and if they take it right. I was fortunate to be in that to have that opportunity a couple times, OK.

But I also saw some people who were who, in their mind, they were stuck for a long time, they were just gonna have to analyse whatever was sent to them. So very reactive. Yeah. And that's probably the reality of some, some jobs also. Yeah, right. If it has to be done, it sometimes has to be done. But. Yeah, that makes sense. Is that where you want to be? Yeah, that's a good question.

Before we get into kind of data science in the field nowadays, you referred you to me you I had her in a previous conversation. We talked a lot about kind of what does TomTom do with generative AI? How much are you involved or how much are data scientists involved in kind of the Gen. AI stuff there or is that data engineering? From the start. From the start, yeah, yeah, yeah. So I was the first full time employee on that story that she talked about.

And that's cool man, what a year, what a year. So for two years before that I was like super excited about whatever Transformer models and all the advances in NLP trying to sell this everywhere, all the time going, look, this problem can be solved with that. And then it's got an interface and everyone realized what I meant. That's all you needed to do. Exactly.

So then I mean the really the opportunities were created because I felt like I was allowed to go invest some time into this because of her efforts. Yeah. And I got the opportunity to do that full time. So that was, yeah, that's great, that's great. So yeah, the meat of it basically is we've been working on building an assistant, a Voice Assistant for cars and you can imagine for data scientists, that's an excellent place to be. I can see that, yeah. And you're still working on that?

Yeah. Because I feel like that's never gonna be done. Done. No. So you have to right now. I'm trying to figure out what my what's my place in that. OK, yeah, yeah. And are you more or do you get

more energy out of figuring out how things work, or why things work, or what needs to be done and then being hands off with that work? Or how much are you involved in the nitty gritty? Let's say detail 'cause I know you have a mathematical background as well. Yeah, well, what? What? I said before. Building stuff. That's the one. Yeah, yeah, yeah. Yeah. Yeah, although I'm starting to care less who actually writes the code. OK, but I do. Yeah, that shit.

And of course, trying out the new techniques, staying up to date with what's possible. Mm hmm, because that allows you to react and spot the opportunities.

Yeah, exactly. So you know now, after there's been this huge conference at CES where every every car manufacturer is showing off that they are considering putting AI in their cars, you know, and how you know to to to stand there and show that you need to actually be aware of the opportunity, you know, one year before because someone needs to build something. Yeah, exactly. That was cool. So for that you have to read the papers, know what's happening and have the ability to actually

build. So it's not just something you're thinking, it's actually look, here is an example so you can show it to people and get feedback as fast as you can. Yeah, if that makes sense. I see a lot of similarities in there in in myself as well that I I like building like I feel like there's complexity, Try and figure out how how things work, how to get stuff done. It fulfills me, but now more and more I care about the outcomes rather than what I do in there.

And if someone helps me achieve that outcome, if I'm not riding or contributing as much, it still fulfills me. I think for me it's about the outcome of the team, let's say or this product that we're doing. And I'm trying to play around also with my role. That's why I said I'm going to be product manager to see if that part also fulfills me and to try it out because I think that might be interesting. So that's the that's. The Yeah, this will be a very new world for you, I guess.

Yeah, yeah. Management is gonna be so indirect. How you're gonna have the impact probably? That's what I feel like and I've only been there for a few weeks to actually don't even have my laptop yet. And now I'm like, OK, what where is my value right? Cuz it's not on the technical stuff. I have people, I have colleagues that talk about the technical stuff and I can trust them.

So then where is my value? It's more so trying to figure out what we need to do or what the value is, or talking about the domain with the stakeholders we're talking to the end users. Yep, it's like a whole field and I need to pick and choose what to do and I'm trying and do my best. Well, I'm guessing you're gonna have to. There's gonna be way more that you want to do that you can do. I hope so. Probably that problem is gonna be worse than when you were a developer.

I think so. I think so. But also nowadays, instead of from the developer hat trying to advocate for what I think is right, I feel like I have more skin in the game and a bigger authority to say, OK, this is what we're gonna do because of XYZ reason. And I can justify that in such a way that I wanted it justified to me. At least, that's my vision. Good point. That's my vision. Yeah. If I can do justice to that, we'll have to see, yeah. Maybe you promise yourself write

it down. I'm gonna ask feedback exactly this feedback question to a person AB and C in two months. Yeah. And I I did that actually, yeah, I don't. I'm not great with yearly goals. So this year I only have one goal and that's to do more and ask more feedback. So I've I've talked to, I've, I've written this down every month, at least the first month. If I'm on a new Simon, I want feedback and then every X months from there down the line, a full 360.

So everyone involved not just pick and choose who I think is going to give good feedback. I might have done that before. Yeah, yeah, me too. Me too. Fully honest there. But I think it's valuable, especially if I'm in a new role, to kind of learn and grow there. Yeah. So that's that's my that's. A good approach. That's my journey, yeah. Yeah. I I'm curious to hear your

thought on kind of how the, let's say data engineering right even data in general data science field has changed. Because I feel like when I came out of uni data science was like a a buzz term. It was on a lot of articles that said that a few years down the line, data scientists are going to make the most money because that's what attracted me back then. Well, it's honest. Yeah, yeah. So it was definitely on my radar, and I mean, I shared as well due to a combination of

circumstances. I never, I never went down that path. Maybe I lost sight of money somewhere down the road. But in any case, I'm sure things have evolved since then and I I don't know how data scientists work nowadays versus how it used to be. For example, are you in a product team? How much do you collaborate with data engineers or software engineers? Like what is the day-to-day compared to how it was previously? Yeah, I mean, I think the most obvious one is tooling.

Yeah, I mean tooling and cloud. Although, you know, not every. I think a lot of people in tech overestimate how much cloud there is. OK. Yeah. And. What the cloud adoption is, you know, so many enterprises are not, you know, really utilizing it. Yeah, not yet. Or maybe they're utilizing the machines, but no cloud services apart from that, right?

Yeah, I mean just shouting cloud sounds maybe boring, but man, it makes a difference because I remember also projects where, you know, we would actually have to go and make a change in an excel sheet and then maybe e-mail or call someone to request some firewall setting to be changed and then maybe yeah. Yeah, all of that. Instead of you just click, now I can run. You know what I want. Yeah, that's a huge difference. Absolutely.

I think if you're talking whatever, before I became a data a data scientist, the profile that we're looking for often like basically PhD was pretty much a requirement really for a while in the beginning. Yeah, I think that is not the case anymore. OK, And it was the case because basically if you if you wanted to to have an algorithm to use an algorithm, you have to first of all probably implement it from scratch in a pretty low level language probably also.

OK. Yeah. And you needed to be able to do the math there. The the job became a bit easier in that respect, like less academic, Yeah, because there's way more things you can take off the shelf. So that makes, I think, all the difference. So between that, somewhere in the middle I joined. OK, of course.

So I did some of that, you know, implemented from scratch that was more for the research and then after that you know you you'll mostly see how you can maximize, you know the the number of things you finish, OK. So instead of doing one thing, it takes you the entire year now because you have tools you can maybe have multiple shots of having in a year. OK. That's pretty different.

That's cool. Yeah, and I think what what's gonna be relevant for, for people who who are considering joining the field, there's something different now in terms of the role and put the expectations. OK. The title is assigned to all sorts of roles. Of course, like any title. Yeah, exactly. Software engineer, product manager, it's coming in all sorts of things, right. Yeah, you see different profiles within that.

But now the companies that are grow growing up, quote UN quote, they are specialized in the roles a bit more, OK. So there will be well the the data engineer was the first thing to be kind of split off. The difference between you know

being an analyst and a data scientist is becoming a bit the the data scientist is kind of catchall and the analyst is clear. It's more like analysis. So basically usually one off or you do some some BI stuff, the machine learning engineer, the research scientist, OK, yeah, the research engineer. So it the profiles that they're looking that you know, companies are looking for are becoming a little bit clearer than they were a couple years ago.

That's good because they were all in one term and now there's different titles for these things. But still it pays to read the role description and then check on if that is also real, right? Because it happens. I've heard a couple of people that go there go into a company based on some role description and in the end, of course, it's not what they expect, yeah. That always happens. Yeah. Yeah.

That's a hard one. Yeah. Interesting. Yeah, but find out what you enjoy about where you are anyway, Then that informs the decision of what you're gonna do next. Yeah, maybe in the same place. Maybe somewhere else? So the role in and of its own is less of a catch all and it's getting more specialist. Is that what you're saying? Like, yeah. I think so. They're taking bits and pieces and that's then kind of those responsibilities fall under a different title or a different

umbrella. Yeah, that makes sense. I. Think, although I think in the beginning probably the entry level job probably still reads Data Scientist or Data Engineer. Yeah, I don't think. As far as I know, I don't think there are gonna be entry level machine Learning engineer positions for example. OK, maybe there are, yeah. Yeah, I don't know. I've I have not looked into this for a long, long time, obviously because I am not in that level yet, nor mostly in the data

field. But I do get questions with regards to software engineering. For example, if I were to start out now, would I do for example a PhD because there's pros and cons there? Would I do even a master? Some people are questioning that or do I need a bachelor's people go even in front of that. And for software engineering I have more of kind of a clearview there I think here and there. Or what you can do portfolio wise, or how you can distinguish yourself. But I have no clue when it comes

to the data sign of things. Is it comparable? You think, Is it OK? Build up a portfolio, show what you can do, Is it similar to that? Or how can you distinguish yourself when you're going, let's say, for those entry level data science positions or data engineering positions? Yeah, well, PhD is not a necessity, OK? I don't. I don't think so. Not anymore. Not anymore. Unless you're looking for a research position. And even then, probably if you got the papers but not the PhD. Yeah, OK.

Yeah, I know you can qualify in all sorts of ways, right? So the and and if it's your first job, all you're trying to do with all of those little experiences and it's just qualify yourself usually just to get an interview basically that whole list, the CV, the whole point is that they're OK to talk

to you. And then probably there's you know, in some decision making a little bit based on the credentials, but will probably be mostly the impression you give and the answers you give in the interview.

Would you say then you need a portfolio to get that seat at the table or what's your experience there? So that's what I assumed when I got into it. OK, so I started, you know, whatever, doing half a project. And then I thought, you know what? In the meantime, let's also go to all the meet ups, OK? And that's what got me the first job.

The meet ups. Yeah. So while I was doing this portfolio stuff, I was like, OK, let's see what the people who are already in the field are actually talking about and thinking instead of my impression here from outside and the Internet. And maybe, you know, maybe doesn't match reality. Yeah, I love that. Started going, you know, just

talk to people you know. Have some questions ready, Because you might, you might be, you know, depending on how you are, you might be too embarrassed, or you might be nervous. Or like, you know, what did you enjoy doing today? You know, at your job, or if you weren't working where you are, where. Would you be exactly? Yeah. So there's no cop out there. I'm like, by the way, we're

hiring. Yeah, Yeah. I think that's that is super smart and you can do that within any tech related field, whether you wanna be a cloud engineer or software engineer, anything data related, going to meet UPS, meeting the people there. I think that's a really, really good thing you did. Was it consciously or were you just interested in meeting people in general? Yeah, somewhat conscious. Oh, really somewhat conscious.

That's good. Yeah. Although I thought it was gonna be the portfolios that did the. Of course. Portfolio that did it. Yeah, it wasn't. No. It took me by surprise, kind of. You know, it's like just apparently some impression I left on someone meant an e-mail from some person to another, and then all of a sudden there was some interview. Now then, you have to pass it, of course. Exactly. So there's some technical assessment and whatnot, and it's

the first job. So, you know, you also have to know what to expect. Yeah. Now, I also made sure that the that the the places I wanted to get into were not the first interviews I took. OK. Because chances are the very first interview, it's not gonna be your best one. No, you have to experience. That, yeah. Yeah, yeah, that makes sense. I had that as well. I interviewed a lot and I did do it for the positions I wanted to get into, so I didn't give myself some time to ramp up.

But yeah, at some point you get better at interviewing and then that that's what gets you the job. Yeah, so I get that. I think the meetups thing I I don't know what your opinion is, but I think that you can still do that because there's still a lot of meetups, although they're not as much as they used to. Be OK because I I've been looking lately and it's my impression is it's been drying up so it's moved online so that they can have bigger audiences.

Yeah, probably. San Francisco still has that thing going on, but even for, you know, whatever, a medium sized tech hub like Amsterdam, I think it's or it's not. Is dry Dry. Yeah, kind of. I see so. I don't know how it used to be, but I see like a few meet ups, A couple meet ups, let's say a month or maybe one a week of things that are interested in the software domain. Maybe it's because I'm at CBA and we also host the meetings, let's say, but that's why I I

see them more. I don't go to all of them. I've been to 1 Kotlin related one, even though I don't even program in Kotlin just to see what the vibe was. And I did enjoy kind of talking to people and being like, so where do you guys work? What do you do? Like, that's a good vibe, but since you're saying it's kind of

drying up and it's moved online, how would you then distinguish yourself nowadays? Well, still go. If there is something go, yeah, I think you're reaching out to people. Maybe not necessarily the if you don't know anyone, maybe you, you likely know someone who knows someone. Just asking. You know, like, do you know anyone who works in that in that field who'll be willing to talk about it? I mean, I I didn't know where to get started either.

I probably didn't even know about meetups, but you know, you got to start somewhere. Starting. Yeah. And it's very easy to fall into the trap of just being behind your laptop and reading all the blog posts, especially all of the blog posts written from a perspective that's maybe not relevant, like it's in a different city or someone had a different educational background, or right. You have to find your path.

Exactly. That's the, I think that's the hardest one and they're always going to be unique. I feel like as well that it's going to be similarities just by virtue of being unique as a person, your path is also going to be unique, I feel like. But many of them start out in similar ways, right? So there's some, there's some beaten tracks you can try to take, like the consultancies, right? There's different flavors of those, but that will expose you to a lot of different situations

very quickly. Probably, Yeah. Just throw you too hard stuff lightly, You know, there's trainee ships and then there's everything else, which is actually the majority. Exactly, Yeah. Would you still, Because I I had an opinion on traineeships when I started out. I didn't like the commitment of, let's say a year and a half or two years, even one year of being stuck. And if I wanted to get out, I would have to pay a fee to get out.

Oh, really? Yeah. Yeah. Because of that educational ramp up. That's what trainee ships. That's my connotation with traineeships. Yeah. Well, my yeah. So when I joined INGI think I had a choice between joining the traineeship or going straight into the job. OK, just give me the job. Yeah, if the point of the traineeship is to get me that job. They give me the job, yeah. Yeah. And I think also the pay was better. Of course, they don't have to reserve, right, your your

education costs. Yeah, yeah. So do you. Think, although I think I I negotiated as a must that it would get some courses with GO data-driven. Your subsidiary, yeah. Yeah, that's. Really nice. That was really important to me. I was like, OK, whatever the first one is, it should maximize the situations I'm exposed to and learning whatever the first thing is.

So yeah, that was my mechanism. Like, OK, gotta get that training and some kind of gig where I can do different projects, not just work on one super long report that's gonna take three years to write or something like that. Yeah, gain as much experience as you can and different perspectives. Yeah, because that way you'll you'll run into things that you realize actually drain your

energy. And maybe you'll you'll find some part of some industry that you can just throw yourself, completely immerse yourself into. Yeah, exactly. Interesting when we're talking

about portfolios, cuz I have an idea of what I would build, let's say from a software engineering point of view. But when it comes to data and data scientists, what would you put in a portfolio to then also have, let's say as a resume? Because I think connections are gonna be valuable. But a portfolio, I wouldn't say it's not valuable. No, I think what a portfolio. Well, most portfolios are gonna

have. Most portfolios are not going to contain some kind of GitHub project that has like 20,000 stars, right? And even if it does, someone has to actually see it. Exactly. So if you're applying and no one clicks on the link of your profile, then it was actually yeah. So really, I think what people are looking for is that it gives a good general impression, like this person did something, but it's also not a necessity in my opinion.

Having a few samples of code that you wrote, I think that's the main value so people can check it out before having to commit to inviting you and having some coding interview, anything that you know. If there is any academic work you did and you are allowed to open source it, why not do that? No brainer I think, and that's it. I mean, don't take it way too serious. In my opinion, OK. I think it will have, you know, diminishing returns if you like, if you put all your energy in

that and it's not. And no energy is going into other ways of being noticed or or standing up because I think right now the job market is a bit harder than when I got in. So you have to stand out a bit more. I agree with that. And real world experience beats, you know, theoretical experience. So internships probably help, especially if you have something

tangible at the end, right? If you can say I did this thing, especially if then the company writes an article about it or something like that, yeah, it's like external proof point. Exactly. You can point to it, yeah.

I like what you said when it comes to not going too deep on one of the aspects, right? Cuz you need a little bit of a lot of things, yeah. To get the job at the end you need maybe the connections because they can put in a good word or they know you, right. It's still, it's still people interacting. People need to want you and their team. You need to show what you've done, but not to such a degree that you're only gonna do that because otherwise it. Loses its purpose?

Yeah, you'll be pigeonholed. Exactly. And you need to be visible and your resume needs to be up to par to get that first initial conversation at the table and then you're at the table and then you have the interview. Is the interview like? How much can you prepare for the

interview? And I don't know if it differs from software engineering to data science, but in software engineering you have the typical OK, let's do a pair programming session, or you do a take home assignment and then you present. Or you do nothing technical and you just talk about how how you would do within a team for example to gauge your experience. I I've seen both of these styles, yeah. And honestly, there was.

Once I even rejected the job offer because the interview process gave me a bad impression. I was like if if if someone can get that position without external validation or doing any kind of technical assessment, yeah I'm not sure if I wanna work there. Yeah, that speaks volumes. Yeah, yeah, yeah. Well, yeah, you're gonna have to be able to answer some technical questions, of course, but don't think then you need to be able to answer all of them or from all different areas.

Just know what you know, right. So if you know that you took a course on something, you know, have, you know, your, your one paper sized summary of it, you know you know it. Yeah, so that when it comes up, you know they're they're most likely to gonna ask questions about stuff that you are claiming to know. Yeah, and especially for an entry level position I don't think anyone is gonna blame you for. Not knowing what you don't know.

Yeah, yeah, I think that's fair. It's maybe it's because the job market is quite harsh that people look at it as kind of this big hill that they need to climb to get their first

position. And I don't know if they're, they're doing exactly what we mentioned in as in not just pigeonholing yourself in a portfolio and actually be visible on on all sides of things and do your homework right, what you've put in that you know you should actually know because that's what they're going to ask about. They're not going to ask you what you don't know or what you said you didn't know because then why, why would they hire

you? They're going to know more talking about what you said you know, and learn from that than from the unknowns there. So, yeah, I don't know if it's

the market or the combination, but I do think it's still very much doable to prepare yourself to the. Preparation is worth it. I mean, of course it depends on what you're putting your energy into. Preparing for techno, technical interviews, I mean, it helps you work on fundamentals. It's also easy to go completely overboard with it. It should just be appropriate for what you're aiming for, I think.

Yeah, yeah, I agree. So something I noticed with people who are just getting started is that they and I maybe had the same issue a bit. If you're coming out of university and you were a good student, you're not so used to like not, you know, many of them didn't have so many failing grades, right? And maybe being rejected. Is a fail. Is a fail. Yeah, it can feel like one, yeah.

Yeah, whereas you know if you see it as a numbers game instead, or like the process of improving and sometimes you have a shot, sometimes you don't, but you need to show up to know. Exactly. Yeah. Yeah. I get what you're saying there. It's my brother was looking for a job. He landed one. But the whole period leading up to it, it was, yeah, disappointment, let's say, or a rejection. Or not even getting invited a seat at the table and not getting the feedback.

Like those are all contributing. To the feeling. Yeah, yeah, it's. And I could see it like I could see. The incentives just don't make sense. There is no upside almost. Or maybe I'm misunderstanding the situation, but I don't think there's much of an upside for a company no to provide very honest feedback to the applicants who didn't make it. I get it, but it's it's a way of giving back. I would say that's the only incentive, right? It it feels like the right thing to do.

Yeah, but I I get that time wise, crunch time you don't have to do it. So it's easier not to do it. And also he he perceived for example things in the interview rounds that he would say like is this a company I wanna work for? Is this a company I wanna work with? Just by virtue of recruitment recruiters saying what they were gonna do, being on time for certain sessions, calling when they said they would, things get sloppy, and also when the market is worse. It gets when the market is

worse, it will get sloppier. Exactly, yeah, that's what if they. Can get away. Why not? Why not? Because then just go to someone else. But then this, this notion of kind of disappointment or this notion of a failure is again a kind of combination of circumstances, right? It's not just the person that's applying, it's the company, it's their internal organization. Maybe they don't have budget. Anymore not even know how many people applied for that one position.

Exactly. I mean, I remember I think at some point I was involved in some funnel. I mean on the hiring side, a funnel that had like for every position there were like 300 CDs. Wow, that's crazy. Yeah. Makes you know if you didn't get invited, it might not be you. No, no, it doesn't have to be you. I mean, it could literally be the software that's scanning the CV. Yeah, you might not. Your name might not even have shown up at the hiring managers table. Exactly, yeah.

And that's by the way why the networking part? You know your name can end up in a different pile. Yeah, on top of the pile and. The smaller the companies are, the higher your chances are that stuff like that can happen because they they don't see that as a problem. It's just how it works. Yeah, yeah. You know, a guy. Yeah, exactly. That's a great point. I'm wondering, let's say when

you land the job, how can you make sure that your growth doesn't stagnate? You mentioned a few things, and you mentioned one of the things that stuck with me is actually making sure you can finish things and hopefully multiple things in the same year instead of just one long trajectory which might not finish. Well, you don't have to be sure you can finish it. No, no. But it helps, yeah. Right. A more a more certain percentage.

Let's say, yeah. Or if even if something feels futile in terms of the final impact, well, someone made that decision to undertake that probably it wasn't you decided that, you know, is there something you can get out of it anyway, You know, there are some projects which whatever, to some of the people who are in it feel like this very, like high likelihood is going to fail. You're gonna probably at least get a conference talk out of it. Yeah, there's so many, you know.

That's the. That's the thing. Yeah, I think so, really. Yeah. OK. Because I wouldn't say that would be but. You can. You know, if you have something short that fails, it's a good story. You know, if you have a really long odyssey and it fails, then you feel like. You're that's wasted. Exactly. Yeah. And then you can start again. But is that, is that how you grow really quickly, let's say as a data scientist is by doing more of those in a shorter

amount of time? Or how do you make sure you're you're growing as fast as possible? Yeah. So I at the time, getting started really enjoyed being part of two different things at the same time. But I think that has to suit you. Yeah, because when when there was a downtime in one thing, it's probably compensated by happy times in the other one. OK. I kind of had to do that also because they were more projects with a lot of moving parts and international and politics.

So we had to, you know, there were quiet times. OK. Yeah. And I'm not very good with those. Quite. You gotta be busy, yeah. Yeah, then I Because then I start to think about why is it quiet? What do I need to push? What do I do now And then. And then I start to go outside of my sphere of influence with my worries, and that's not good. Yeah, it can be bad. Yeah, yeah. Yeah. No.

So how? How to maximize the learning or take off, you know, so the variant of being being two things at once is like, you know, expose yourself to different situations. Because that way you will get the most learning in terms of not just the content but about what vibes with you. Maybe you don't like writing an an analysis. You know, if you if you like analysis but you don't like writing, writing reports is

going to make you unhappy. That's a good thing to come out of. You know, if that took you half a year to learn, yeah, because of project, that's a great outcome. Yeah, take that with you. Take that with you. Yeah, exactly. Yeah. Yeah, you can vary the the technologies and the methods in the beginning. Just expose yourself to, you know, different methodologies, even families of methodologies. Not everything is deep learning,

for instance. Yeah. And the people that you work with are gonna have the huge impact, I think. Yeah, because just like you were saying, you asked for feedback a lot in the beginning of a new venture. The outside perspective, People from the outside sometimes pick up way quicker on some aspects of what you're doing, then then you can yourself. Of course, there's some other things you notice really important to be around people who have more experience,

especially in the beginning. Yeah. And that's, I think what kind of what I described when we're just starting out talking, that that's kind of what went wrong. The first project I did professionally, Yeah, because I went in with someone and they left and I decided to stay on for too long. Exactly. Yeah. That's not a good sign.

Yeah, I like that a lot. And it's something I I think always it would be good, even if you're to the highest on the totem pole, hiring someone that you can learn from, but that is better in some aspects is always gonna be good. It's gonna make you better professional, I feel like. Yeah. So I would say always try and work with people that you can learn from. Yeah, yeah, that's great. Yeah. But in the. So yes. And in the beginning, the focus, I think, is more about also the

content. How do you conduct yourself as a professional? Maybe other people spotting things about you you didn't know, like what strengths you have? Yeah, maybe you're too humble, or maybe you don't take enough risks. Or yeah, that's the type of things people will talk about in the beginning. But in the beginning also, of course, the actual content. Like how do you? How do you do the thing? How do I work? Exactly. Yeah, that's a great point.

When we're talking about feedback, is there something that really stuck with you kind of early on that someone mentioned? Yeah, there was one piece of feedback someone gave me. Well, I think the second year, that just really stuck with me because I was really unwilling to take on to like accept certain risks sometimes, OK. Whereas actually the impact of that risk, it wasn't gonna come back to me, but I just wasn't comfortable with it.

And then someone confronted me with that and goes like, OK, so you keep saying no to XYZ. Yeah, probably the outcome is gonna be better if you say yes or really leave. OK, well. So you have to make a stand. You can't be in between all the time you. Have to make a decision. You have to make a decision, yeah, Yeah. Even if you disagree with something, you're gonna have to, you know, commit to the decision or leave. Well, I think. Politicians are best at that, right? So. I I think so.

Is that did you commit or did you leave? Because that's an interesting one. I committed there and the outcome, yeah, the outcome is like a very good friend now. That's not. The person who gave the feedback, but someone who I met through that project and then everything turned out all right. It's a, it's a. It's a fine line on the CV, but I know now that that area. I'm not sure if I would touch it again. Yeah, exactly. It's a learning either way, but I like that it had a great

outcome. Yeah, in that way. Well, maybe rationalizing it a bit. After Could be, yeah, yeah, could be. I think that's also better to then dwell on the positive instead of the negative. In that way. Yeah, for sure, Yeah. When it comes to cuz this is

kind of my final thought. I've never worked with data scientists in the same team. Maybe that's because I I didn't touch as much. I haven't touched as much data components. But do data scientists work, let's say together with software engineers or data engineers in the same product team, same way of working? Or is it more isolated data scientists in a team in and of its own? What have you experienced or or how is the way of working? I've seen a bit of both.

OK. This depends a bit on just the company, what they're trying to accomplish with it. Yeah, there's also the two kinds of applying data science. So one is informing how the product should be built. That's more like analyst in a way. I think product managers rely heavily on this, on this

profile. I think booking.com in the Netherlands is a very good example of a company that rely where the product is really informed by what those data scientists who act like the analyst for that product. It drives the product direction. The other side of it is maybe when you put statistical methods in your product, that's a very different kind of. Your day will look very different if you're of that kind, type A or type B. Exactly, yeah. These are actually totally different jobs. Almost.

Yeah, it sounds like it. Yeah, yeah. I enjoy the one where the statistics and the math and the whatever the methodologies are, they end up in the product. I find that most exciting, but can be anywhere in the whole how it works. Like forecasting for energy companies, right? So they gotta know all right chances that we need this much energy between that and that time. What's the 99th percentile of the demand? Exactly, yeah. Then the product itself is based on stats, yeah?

Yeah. I was thinking about the latter. So exactly what you also your preference is where it's part of the product, Yeah. How do you then make sure that it becomes a good part of the product or that it gets integrated properly in that way? Do you do that together with the rest of the product team or have you seen that way of working? You'll have to do it together, of course, Yeah, right. In that case it comes down to project management, partially, I

think. So spotting the opportunity is one thing, but ideas are basically free. Everyone can have ideas all day long. Yeah, I've had to kick start these little things very often and you've got to be willing to kick to kill them also. So maybe set yourself a day to where you think, OK, I will try to have this thing, you know if if I see this then it will be right. You might not see that thing, you might see something else. But work towards that.

You're probably getting to justify you spending time on it. You're probably going to have to convince the people around you and they're going to give you input probably on what will make it a success or not. But when you're setting up, when you have this idea, I would say, you know, really write some clean document without distractions. This is what I'm going to try because of this. I know it's a success. When I see ABC, what am I trying

to learn in the first phase? Because in the the first thing you're going to build to get towards that goal, it's very unlikely to be the thing that ends up in production. OK. Yeah. At least I've seen usually. Usually, unless a very similar thing had been built, you know, a couple months before, that it gets scrapped and rebuilt. Especially if it was built by just the data scientist. OK. Yeah. Simply because of the engineering skills involved in building something for

production. I think we talked about tooling before. It's changing a bit because if you're in the right platform, yeah, yeah, there's less engineering work to do. Yes, more established. Yeah, I know. But setting up the project right is a huge part of it. And then keeping momentum, it sounds like every week you can, you know what progress you made. Exactly. That's kinda. At least for me that helps. It sounds like this. I see experimentation as a skill, and it seems like it's

very valuable here. Also to let go and not be like people always say. You're not. You're not your code, You're not the code that you write. Which also means you need to be able to let it go and get thrown in the trash, and that'll be redone in a. I'm happy I'm not the code. That exactly, yeah. But it feels like that is a skill you need to develop or would be good to develop that experimentation mindset. Yeah. Is that?

Yes. Yes. And one I mean especially if you have a leadership position in in

in this whole story, you're gonna have to make clear that it is experimentation. And so to give them a real to give the people around your realistic expectations of OK, we're going to try 10 things this year. If one of them pays off, I'm happy maybe I don't know what your number is, right. It's going to be it's it's it has to be related to, you know, have we done this before. If we haven't chances that we find you know the right things

are lower. Do we know that in the same industry, people tried this thing and it paid off? There's all these factors like this, and sometimes you just have the wrong idea and it will take a while to find that out. There is one little analysis you can do beforehand that sometimes kills your project before you start, and that's a great outcome, I think, because it means you didn't even have to do the experiment. Yeah. Just an argument about maximum potential impact.

If the total impact it could ever have, it's not appropriate for the input you expect. I don't see the point. There's probably other opportunities because if this is the last opportunity, it's all you can do in the world, then you do that, but. Yeah, otherwise you have options. Yeah, that makes a lot of sense. Do this at the beginning. This is what you do day one,

yeah? And also because if you identify the total impact it could have, it can help you get excited about doing it. Yeah, and goodbye in as well, yeah. And that one. Yeah, it's that's funny because that's probably one of the most similar things I can relate to. If there's a story on the board that I don't agree with or that I think we shouldn't build, I'd rather not build it and have that conversation. Then start building it and in actuality figure out it's not

being used or it's a nuisance. Let's say that I think in general, with anything you do might be a good skill to develop, figure out if this is actually what we need to do versus the other things we can do, 'cause there's always gonna be options. Yeah, yeah, I. Think this is gonna be a big topic for you and as a product manager. For product managers, for sure, yeah. That's why I really look forward to it. Yeah, this was a blast, man.

Marcy, this was a lot of fun talking about kind of this career journey for people that get into the field nowadays. Also learning from how you did it as well as how you've grown and how you would grow, kind of starting out now with the technologies that are available, looking at the people, the tools and what needs to be done ideal wise to production, Is there anything that's still missing that you'd like to add? No, I love this. Yeah. Yeah, absolutely. Good stuff then.

Thank you so much for coming on and we'll we'll round it off here. Thank you for listening. I'm going to put all Massey socials in the description below, check them out, let them know you came from our show. And with that being said, thanks again for listening. We'll see you on the next one.