TGHH 37: Data Scientist AMA

Welcome back to our career deep dive series! Ever wondered what a career as a Data Scientist is like? Join us this week as we interview our friend and guest Meera on life in this exciting hot new field. We'll discuss how to start a career in data science, the day to day, and the highlights and lowlights.

Resources mentioned in this episode:
Masters of Scale podcast
How I Built This podcast

Let's stay connected! IG: @techgirlhappyhour

Transcript

0:09

Hi. Hello, and welcome back to another episode of tech girl. Happy hour. We are your hosts. I'm Mel and I'm Horace. And today we are joined by a special guest. Would you like to introduce yourself? Yeah. Hi, I'm Meera. I'm a data scientist and I also work with Marissa on the same team. I'm from California. Um, I've been up in Seattle for a few years now and I love it. It's definitely my vibe, all the hiking, the brewery scene. All the outdoors and nature. That's um, that's what I love to do in my free time. Awesome. We're so excited to have you join us on the show. Yes. Thank you for joining us on the podcast tonight. Before we jump into things, the most important, the most pressing question, what are you drinking? Meera? I am drinking a hazy IPA, which is if you know me, it's very on brand. It's from, I think, stone brewing. It's from somewhere, probably on the west coast. We joke that Meera is like the most west coast person. We know, like I would be so sad if she left the west coast, she like drives a Subaru and wears Burks and goes hiking and drinks, beer very close. So you definitely sound very on brand to be friends with Marissa. Very similar. That's why we get along now. What are you drinking? I tonight have a bottle of this broadband Vino bear day. Oh, yes. I think if a Vino rare day, a different brand on the show before, and you know what I was at world market the other day, so random, but they have for most of the world market, a local market shop there there's no, there's no focus. Like there's no like product markets. It's just random, but I think that's their whole brand. That's the point of it. But I agree when I first went there, I thought it was only a home store. And then I went in and all of a sudden there was like, Food and alcohol. And I was like, whoa, wait, I did not know about this. And they actually have a really cool, um, like drink selection. So this one, I Googled it and it sounded like it had good reviews. So I got in, you know what? It's pretty good. Wow. Neat. The more, you know, how about you, Marissa? What do you have? So to make up for my lack of drinking last episode, I. And got some fancy craft cider. Um, I've to actually, and you can see me, these bottles are huge, but I'm almost done my first glass of the first one. It's a Eagle Mount ginger peach cider, um, tastes like Caboolture. Honestly, it tastes like a sweet kombucha. Nice effort vessel. A little bit too sweet. Um, so hopefully the next one's more dry and uh, you'll you might hear me pop the next one. As I finished my drink, this one's Dragon's head Columbia crab, apple cider. So I think they're both local to Washington. Yeah. Nice. I'm very fancy. Yeah. You know, sometimes you've got to treat yourself like yeah, love it. All right. So this week we have another guest joining for episode, um, as part of our AMA series. So back then when we did the software engineer AMA and the PM AMA, we got pretty good feedback on Insta and also got some good listening stats on that. Like our engagement metrics are pretty high. Um, so, uh, speaking of engagement metrics, we're going to dig into life as a data scientist with Meera. And she's going to answer all your questions about. Well, she does getting a job in the field. Uh, yeah. So why don't we start with, uh, Meera, can you tell us just about your job? What do you do? Um, yeah, so data science, I guess, overall the idea is like understanding or, you know, observing all the usage data, um, behavior like user data of any product and manipulating it, modeling it, looking at it to try to find some insight about the users or. In a way that you can improve it. Um, so that's kind of a high picture of what like data science is, but there are like three main parts to it. One part is experimentation. So AB testing, um, you know, I would say like every, any two people, if you pull up the Facebook app on their phone, No two apps look the same because there's some difference between the two of them. And that's Facebook AB testing that different change. Try to see like which one is doing better with like the customer base. So we do a lot of experimentation. We do a lot of like bigger picture, um, strategic problem solving. So, you know, understanding usage data from the product and trying to determine is the, um, is our strategy or roadmap going in the right direction. Are the designs for some certain feature, what they should be, or like, how should they change? And then the third part of it are we answer a lot of ad hoc questions from anybody else on the team. So PMs or designers or leadership ask questions, like how many users are doing X, Y, and Z, or, um, why is our performance like a, B and C? And we try to help them answer questions like that. That's what data science, this is awesome. That was a good overview. And I feel like these series also helped me to like understand more and like, better about the other people and other roles that I work with at work. So. Awesome. Um, what, like kind of education background would you say is necessary for data science and what did you do in terms of school? I think the most interesting part is like the different sorts of people that I meet in data science. Um, I've met data scientists who came from an econ background, or they were a teacher in their past life. Um, people from finance or obviously people from like maths or stats, or even computer science, because I mean, one, you can use data science in so many different disciplines. Um, and people kind of come to it from all their different interests that. And they apply it in that same field. Um, but personally I did a computer science, um, as a major and I did, I got a math minor also, and that's probably where like the stats and math interest came from because there's a lot of stats and data science. Um, and I did not expect to go into data science at all. I don't know if I knew what it was when I went through college. Like it wasn't a degree or major option when we went to college, but it is now. I think I slowly discovered it, like during my last internship or, you know, my first year of work and realized I clicked with that a lot more than I did with regular suite or coding. Hmm. Awesome. Yeah, that was actually going to be like my followup on, I feel like no. Undergrad program trains you for data science that well, like, I guess you could have like a stat be a stats major, but it seems like one of those like kind of niche tech fields that people just end up in like same with product management. Right. So can you tell us a little bit more about that journey? Like you were doing software, like, were you always certain that software was going to be your career or like was there, what was the journey of deciding on. Uh, that's a tough question. Um, I truly cannot pinpoint when I decided to do data science or when I learned that was a thing. Um, I joined like the data team of my company when I joined a full time as a. Because everybody on that data team, we were just all Swedes and we were doing everything under the hood of like data. So data engineering, data pipelining, and some like data analytics. And that's probably when I discovered, oh, data data science is like an actual field aside from sweet. So they're, you know, dev teams and PMs. And then there's also data teams on, um, in software companies. And so I did a bunch of research into like, what does. And Airbnb data science team look like, what does a twisted a team look like? So all of these like big tech companies, big data companies that I knew had really good, like data foundations, just to understand what those teams look like, what sorts of problems they were solving, why they were solving it and how I could bring those lessons back to my own team, because we were a pretty neat like nascent data team. So I think just like learning about it, kind of spurred that interest in it and. I realized it just sort of clicked that I really enjoy that sort of work. And personally, I felt very grateful because I was like, finally, I can, this is the sort of coding. Like the scripting coding is what I like. And I'm better at versus the hardcore, you know, client side coding or building a product. And so I realized I really liked the scripting. I really like learning about the data and doing something cool with it. And then translating those insights via a story or a presentation or something like that. So I loved to find out that this whole discipline existed. Oh my gosh. I love that. Okay. I feel like you've given us such a good overview of like, kind of what the job in the field is all about, but take us even more into your life and tell us what a typical day would be like. Um, for a data scientist. Yeah. A typical day is like, it could definitely vary be two week. But I think the key parts are a lot of times we're working with multiple different feature teams in a more like consulting manner, I think. Um, and so that could be working with, you know, PM or design to figure out what success metrics you should be coming up with for that. Um, and so that's a consulting role. I say that because we're not the ones coming up with the actual answers or the metrics, and then giving it back to the PM. We work with them and sort of try to guide them or instruct them, um, into coming up with those metrics themselves because they are the product specialist or the feature specialist. And they know the most about that. So we do, I guess, quote unquote, like consulting in that point of. Um, another part is, like I said before, like we have the bigger picture or like longer-term data science projects that we all do. Um, and so we each have probably like one that we're doing over the course of the quarter. And so that's why every week could be a different journey. Like one week you could be the whole, you could be doing the brainstorming part of it. And another week you could be working on gathering the data, cleaning the data the next week you could be like working on modeling or visualization. And following you could be working on the reporting or presenting aspects of it or the storytelling aspect of it. Um, so I'd say those are like two main parts of it. And then the third part over day to day is always working on some ads. Question that was asked of us either by like somebody in a product team or leadership for, you know, an ad hoc product review that's coming up. So there's always that, yes, I am definitely guilty for ad hoc questions for Meera. I believe I've done that this week. Part of the job. Oops, sorry. Um, okay. That's really cool. So you do a ton of things. Uh, Crazy amount of variance in the job. Can you fill us in a bit more on like, what is the most exciting part of the job? And like, what's the least exciting part. I would say the most exciting is maybe it's a personal experience and maybe it's unique to our team. I don't know, but I really like being able to influence either, you know, the, a decision, um, in a unique way and. I feel like at least now data is still like an up and coming field that like, some people understand, some people don't. And so anybody who's not in the data discipline or a data scientist might not know enough to be an armchair expert at it. So they're not going to like push back a ton or think. Backseat drive it. They're kind of going to just listen to what you have to say. And like, we got to be the experts in the room, um, even like, you know, more senior people. So that's something that I find very cool. So we, you know, have a lot of sway or we could have a lot of sway. Um, and the second thing is I really love to like, learn something about your customers or users or whatever product you're learning. That's not that wasn't an obvious thing just by looking at the product or observing somebody. Um, it's sort of like you're digging something out of Pandora's box that like you just didn't know before. So those are the two things that are super cool to me. Oh, okay. So, and then what's not cool. Like what DOL stuff that people should know about. I think the most frustrating part is you don't always know all the data and. So figuring out how to be confident in a message we're delivering. But knowing that we don't know all the data, it's like an NP hard problem. Like, you know, you don't know something, but you don't know how much you don't know. So that's kind of the one frustrating part. And then going off of that as no data is ever perfect or clean, right. You always have missing data or sometimes you don't have the data that you actually really want. Those are the more real aspects of it. So it sounds to me like you kind of, you know, you've found yourself in this role that you weren't initially planning to. So you probably, I would imagine went into it with a pretty open mind, but was there anything that surprised you about the job once you started actually working in it? First what surprised me was, like I said earlier, how there are people from all walks of life that can end up as a data scientist. And I love that because they, their insight and their experience is just so vastly different. But then also the fact that there are actually different flavors of data science. And just like that person with that econ background is practicing data sciences, maybe like. It's still in an econ fields or not, or even in a tech field, their unique background makes them a very unique flavor of a data scientist with that like domain expertise. And I love how I feel like every person who's getting into this field has the opportunity to be like their own flavor of a data scientist based off of their interests or even their background or passions or whatever they want to, whatever sort of data they want to learn. You know, there's data in all different fields. So you can beat that flavor of a data scientist. That's actually super interesting. I feel like there's a lot of parallels to draw with PM. They're like obviously very different jobs, but like PM is also a field that you don't really study for. There's people with various backgrounds. My manager has an English degree and everyone's a different flavor PM. So it's very interesting. And it kind of makes me think like, What do you think is like the, like the culture of the data science industry, what's the status of it? And like, where do you see it going? Cause like, I am getting vibes that it's like a very new field that people are trying to navigate. Yeah. Yeah. I think the culture of the data science discipline is definitely like up and coming. It's a, I would say quote unquote, it's a hot new thing. Um, I love that because. I agree with it. I see the space for it that, um, existed before, but like we weren't really focusing on it. And like that space is between lower level, um, data engineering and want to say lower level. Like not, skill-wise just like, you know, data engineering versus, um, like machine learning engineer. There's a spot in the middle for like a data scientist. We're not doing crazy, like operationalizing ML models or anything like that. And we're not building pipelines. We're doing something with all the really interesting data that we have and pulling out insights. So, um, I feel like the status of data science right now is like it's very well respected. Maybe slightly, not as understood as it could be, or as other disciplines are. But I think it's, it's only gonna we're in the big data revolution. Right. And so I think this is just the next step after that. Okay. So, so you mentioned that you started out as a suite and then you were on a data team and then you decided to get more into the explicit role of being a data scientist. Um, what, what advice, or what kind of insight into the process would you give to somebody who. Either already knows that they want to go into this field or maybe they want to try it out, but they're, but they're not sure how to, especially considering that a lot of like formal education and undergrad programs don't really tend to focus on this. Yeah. I would revert to something that Merissa, she mentioned earlier, um, talking about PM skills and like the backgrounds that people come from, um, for data science, I think. What's cool about the role is I consider it, you know, like 60% technical, like, yes, you need to understand the theory of how to build certain models. And I'll also like the application of it need to understand the stats of it and things like that and how to write code, but that's only 60% of it. The other 40% of it is the product side of things. It's the product side is the design side is the stakeholder investment and communication and story to. And those are the seals that you're not taught in school. Just like those are also a lot of the key skills for APM. Those aren't the things you're taught in school. And so understanding the product, what product sends is understanding product sense for your own product. Um, and business acumen are really important. And. While those might sound like buzzwords. Like I think those are things that we need to, or anybody who's interested in a data science discipline. Like they have to do a little bit extra work to go learn about that stuff. They have to be curious about why does what I'm doing with the data? Why does it matter? Like you have to ask the, so what for the data that you just learn something about, it's not just about the insight, it's about what you do with the insight that truly carries it forward. So. Along with the product sense and business document, the storytelling of it, communicating it and knowing how to like manage up or communicate to other disciplines, but also like different levels of employees in your own team is super. And they're not necessarily things that you'll learn in school at all, but there are a lot of really great podcasts or books or articles that you can learn from. And also, you know, managers and other more senior members on your own team or your own company are really great to learn. That makes a lot of sense. Yeah. It's super interesting. I feel like in general, like a lot of the stuff with your work is very much picked up on the job. Um, and it's, it's hard for school to prepare you that well. Um, but yeah, no, I definitely learned something new here. Like I actually didn't expect and didn't know that the people, part of it was such a big part of it, but now like totally makes sense. I guess like, okay. Say someone, maybe they're like an undergrad in college and they're interested and. Done the research and feel like this is something that they want to do. Like what advice would you have for people that are like ready to get those applications in? Like how do you really break into the field? I have a hard time answering that question because I don't know what the undergrad to industry new hire transition looks like for a data scientist. And I'm aware that like data science is now, uh, you know, uh, major that people can get in college. I have no idea what those classes would be. I feel like it would be a version of the classes that we all took as computer scientists, but at least, you know, in the industry, what you are testing on in an interview is probably somewhat similar to what a new hire would go through. But at least for an industry hire it's, you know, one 50. But it's two fifths product sense and case study questions, and then two fists, two fifth stats and business acumen. So already by looking at that interview structure, you can tell that there's a really heavy, like weight put on to product sense and things like that. Those maybe softer skills and stats is a really big thing. I feel like stats is a confusing topic that all of us have to learn in school, but having that like solid understanding of. Those concepts and honestly like details of them, the pros and cons. I think that's really, really important. So for D for, I guess, a college grad at data sciences, hoping to break into the field, I would say focus on your stats concepts, but also think about the product side of things. Think about the application, what you would do with this insight and like the motivation of the stakeholders that you're delivering this insight. And how would you communicate it to them to try to convince them? It's it's definitely like communication and people game at that point. That was extremely insightful. Yeah. I I'm learning so much also. I have to say stats stats is low key, like such a hard math. Like I thought calculus is hard stats. It's like easy. If you just like spaced out and you don't think of why things are the way they are. But like, when I think about why it just, it's just nonsense. Like at least the calculus has like logic stats is just so hard. If you think about why I agree. Whenever I try to work through a problem, it's like the more I think about it, like farther from the solution, I get it just like a cyclical problem. Like it's easy to work with P values, but I've also like, but where did they come from? Like what is it on me? Yeah. I feel like it's like informed ma yeah, you have to. You have to take it into account. So many like real life kind of different variables and like have the awareness to like, even think of them in the first place. Yes. I, I was never good at stats, so I never knew why. So I'm going to, gosh. Yeah, it sounds like there's a lot of different skills that are needed to succeed in this job. Uh, like, and I think all of your insights that you've shared about how multidisciplinary it is, kind of have come from your experience in working in the role. Do you find like for other people, either people who are maybe new to the industry or people who work in other roles, is, are there any common misconceptions that people tend to have? And the biggest one is. When people think data scientists just like crunch numbers and we can get you whatever number that you want. And the answer is we absolutely cannot. We most likely probably can not get you the number that you want either because we just can't or two, because it's not the right question that you're asking. So I think that's the biggest misconception, like. Any question or for a specific number, we will definitely turn the question back on you and ask you why you want it and what are you actually trying to get at? So I feel like data scientists, they're like the Oracles. Like we treat magical entities at work and it's like, I want an answer. Give me the answer, you know, the answers to everything. And yes, many times like off the up, but is this the right thing that you're looking for? And I'm like, Ooh, Sometimes annoying, but yet. Okay. Little known fact about me. I did one internship in like data analysis. And one thing that I always thought was funny was that people would come asking for the same thing, but they would frame the question in different ways to get the answers more, to like how they want to. Yeah. I love that. I think it's so. Funny, when people ask a question, that's so biased. It's like, can you give me the numbers so I can prove this one specific thing. Don't give me any other number. I just need to prove this thing. And it was like, you know, we can't do that, but thank you for asking. That's amazing. Yeah. I mean, it's, I think it's easy for people to use data. Like whether it's you are data or metrics as like a web. For arguing it is. And that's what happens is like my favorite is when there's clearly a little debate going on or a decision to be made.

And one person walks in at 10: 25:32

00 AM asking a certain question is. And like, you know, framing it in the way to help kind of support their case. And then somebody else walks in at 11:00 AM asking the exact opposite thing and like trying to like frame the question to support their case. It must be like, I feel like it gives you a funny insight into like the business workings. It does. It's, it's sort of daunting also in a way, because as much as like, you know, we can joke about that and ask them bias questions. We're getting biased. Even the data that we pull have some sort of bias to it. And so I feel like it's, it's a tough job to deliver an insight with confidence, knowing that is bias in some way, but it's all about like, finding that balance or being able to communicate it, or even like trusting on your teammates to push back and like ask you question that you might not have thought of. So there's always going to be bias in it, unfortunately. All right. So I have to ask because data scientists are very cool and do very cool things. What is the coolest stats thing you've done at work so far? I there's definitely some cool stats or like visualizations that I've made, but the thing that it has excited me the most recently, and I think I've gotten like the most positive feedback from is projects where I joined like quantitative. Modeling with qualitative interviews and qualitative data. So it's been where, you know, we had a really big open-ended open-ended ambiguous question. I did the data part of it. Um, did some modeling, looked at the insights, came up with the presentation. But then, you know, it's still just learning big data and it's still just like learning big trends and you can't just bucket all of your users into like one, two or even three buckets. Every user is very different. So the next step of that project was actually talking to a bunch of different customers to get their actual feedback. So like the results of the data project was the, what of what was happening and then talking to the customers was the why. And so that was just so valuable because it could like fill in the, fill in the color to like what the quantitative modeling output was telling us, and then was able to like really communicate a super strong insight, I guess, um, at the end of like that combined project. And I think the. Um, the really great part of that was it resonated with all the different disciplines. So the analytical output of the project resonated with analytical people and the more softer parts of it. So the feedback and the insights from the customer interviews really resonated with, you know, the maybe designers or user researchers. And so it was great because everybody on the team was really able to understand what the output of this project was. Key takeaways worth. So I'm a huge proponent for qual and quant projects together. Crazy. Whoa, I've never thought of it that way. You know, it's the way you put it with you. Oh, that's interesting. Yeah, no, the way you were describing it, it's like, I can really tell that you, you have developed like a good sense of storytelling. Like, like how, like what, like being able to say, like what parts would help strengthen my case and make like, make it easy for me to communicate to different types of people, the information or the goals that I'm like, trying to reach with this, with this information. So, I mean, Hm. There's, there's obviously there's the stats knowledge. And then you have spoken about like the storytelling skills, the business argument, all these like soft skills. And I feel like we talk about these sometimes and they're, they're so hard to develop. Um, obviously it comes with experience, but it, it needs like a sense of awareness along with the experience. So do you have like any specific resources or like almost. A framework for how you approach new problems or also like how you iteratively improve on yourself in these soft skills. I like that question. And what came to mind when you were asking that is, I feel like I've learned most of my job except for, you know, stats and things like that from podcasts or books, or then other people in other disciplines actually. And there's some really great podcasts. Masters of scale or how I built this and basically product or entrepreneurial podcast, because you learn about product sense and business acumen to use those buzzwords again, learn about how a product can be formed. Um, also like product or business types books, or even like creative or design books. I learned a lot from if you actually read it and try to apply what, you know, their experience to the problems that you might be solving at work. And then third is. I think there's some key people, at least at my company, or even friends or friends that their experience I've learned a lot from. Um, and just observing like how they maybe do their own presentations or what are some things that they tried to keep in mind when they are telling a story? Um, just learning and like being curious and trying to pick up those softer skills from. Disciplines, other than mine, I think is where I've learned the most from such as, you know, a designer or user researcher obviously. PM. Wow. That's super interesting. Huh? Nice. I feel like you're more well-read on the product industry than I am. I got to check out these podcasts. Yeah, there is so interesting. I don't know. I feel like growing up in Silicon valley, I got it. It's like all around me, right? Yeah. That'll do it. Yeah. All right. So talked a lot about your current job and what you do and getting into the field, but we're really curious, like, what do you want to do in the future? What's next for Meera? Is it data science? Is it something else? Like where do you see yourself going? I, uh, I don't know if it's, you know, data science or not data science, but I do know that, like, I just want to work on something and be doing something that's having impact and improving the lives of people. Like personally, I like to work on something that is, it's sort of one way to think of it as like solving third world problems instead of first-world problems. So helping people that in a way that they need it versus a wall in a way that they weren't. So, whether that's as a data scientist or a PM or something like that, I'm very indifferent. I've always thought of kind of trending towards product management, because I thought that was a role where I'd have more say and be able to ship or whatever that is. And I really wanted to come from a technical background and I really like coming from a data background, um, because I believe strong PM's have that data. So, whether it, you know, I actually become a product manager or just go up in the data science field, because I also feel that, you know, data science and a lead in that role can have really big say, I don't know what the answer is. There. I just really like to have like, impact on product from the data side of things. If that answered the question. Oh, that's a really good answer. No, I think that's super cool. Like I think, um, having, I always say that for PM, there's different flavors of PMs and like, The data science flavor is really like wonderful and valuable. And I don't think there's actually a lot of PMs that are like that. Like we pick it up on the job. Um, so I think that that is like, you know, if you ever get into PM, that's a really, really good strength to have. And I feel like your interest in product and product sense in general, like ma well I'm biased, but I think that you would do a great job. That's amazing. Okay. Just to pull it back since we're all making the parallels between all of our industries, you know, you guys have really got me thinking with seeing the like, oh, there's different flavors of being a data. Scientist are different flavors of being a PM. And I'm like, oh, is there like different flavors to being a software engineer? And I think there is to an extent, but I, I do think like the technical skills are. S very concrete and emphasized, but, and I've said this before, but in different words, like in order, like to be a software engineer, there's very clear cut skills that you need to have, but to really Excel and like maybe make it as like a team lead or like just any, anything beyond just the basics. You really do need to bring those soft skills. Um, and having. A sense of a little bit of a sense of product. And I think like a sense of data and like understanding your metrics because you know, you go to, you go to make a change in the code and you want to measure its impact. Well, you need to have an understanding of the data, um, or at least be able to work well with your data scientists. So I think all these kinds of skills are a little bit intertwined and at least in software, it definitely does help. To that next level of being a software engineer, but it's so interesting to hear everyone else's perspectives about like what skills are emphasized in their roles and how that applies back to my domain as well. I think in general, just well-roundedness is so powerful. Like it's just, it's good to be good at many things. It's very good to be good at soft skills, no matter who you are. Have have like product sentence, like any, I think everyone should have product sentence. Like you all work on the product. Yeah. How am I? Gosh, well, this chat tonight was so interesting. Thank you Meera so much for joining. We're so happy that you got to be our second guest on the show. Wow. I'm honored guys. I had so much fun. Thank you for making me like introspect on this stuff, then, you know, verbalize it. We're happy to, it has been a pleasure. Yes, we love. Okay. And with that, we hope everybody learned something new. I know. I definitely did. And, um, yeah, and we hope this also gives you a little bit of an opportunity to be introspective into your own career development and the skills that you take in with you at work. Um, thank you again so much for joining and we will catch you on another couple of weeks.