#205: Nailing the Data Science / Analytics Job Interview with Jay Feng

So, you finally took that recruiter’s call, and then you made it through the initial phone screen. You weren’t really expecting that to happen, but now you’re facing an actual interview! It sounds intense and, yet, you’re not sure what to expect or how to prepare for it. Flash cards with statistical concepts? A crash course in Python? LinkedIn stalking of current employees of the company? Maybe. We asked Jay Feng from Interview Query to join us to discuss strategies and tactics for data scientists and analyst interviews, and we definitely wanted to hire him by the time we were done!

GANs and Other Interesting Items Discussed in the Show

Photo by Ardian Lumi on Unsplash

Episode Transcript

[music]

0:00:05.9 Announcer: Welcome to the Analytics Power Hour. Analytics topics covered conversationally and sometimes with explicit language. Here are your hosts, Moe, Michael, and Tim.

[music]

0:00:22.2 Michael Helbling: Hey, everybody. Welcome to the Analytics Power Hour. This is Episode 205. Getting jobs in data science and analytics is often, I don’t know, a highly individual pursuit. Tim, when you and I were starting in analytics, it wasn’t even necessarily on purpose at first, it was just a thing we both got an opportunity to pursue. Do you remember that?

0:00:43.9 Tim Wilson: Well, I’ve been trying to get out ever since.

0:00:45.8 MH: That’s right. And Moe, you didn’t technically start in analytics, but I think at one point you made a decision to sort of steer the arc of your career towards it. Is that accurate?

0:00:58.0 Moe Kiss: Well, I had limited options and thankfully liked the direction I went in.

0:01:04.0 MH: Okay. I thought it was a more strategic choice like, “Oh, this looks like a good… ”

0:01:08.4 MK: Sure.

0:01:09.0 MH: Yeah.

0:01:09.8 MK: Yeah, it was very strategic. I’m known for my strategic thinking.

0:01:12.5 MH: ‘Cause I always tell people I just Forrest Gump-ed my way into this whole thing.

0:01:17.4 TW: Yeah. It’s a great line to use in an interview when you’re trying to get a job. “I don’t know why I’m doing this.”

0:01:22.5 MH: Yeah, I’ve used that in an interview before actually.

[laughter]

0:01:25.4 TW: Well, as time has passed… It’s long past the time where now people are definitely studying this in school, they’re starting their careers in this field, and there’s tons of areas of mastery. And what’s interesting about job descriptions and the interview process is occasionally the job description might even specify what the skills required are as opposed to maybe like a massive unicorn wish list of every possible thing, which makes it hard to figure out what to do for a job interview or how to prepare and doing that interview dance can be stressful. Preparing effectively is sort of a strange combination of knowing the basic steps while also showing off all your best moves and that’s frankly just really daunting.

0:02:09.0 TW: I know I have not interviewed for anything in a while, and so we needed a guest, somebody closer to the action. Jay Feng is the CEO of Interview Query, a company designed to help data scientists get their next job. He’s held data science roles at companies like Nextdoor, Monster and others, and today, he’s our guest. Welcome to the show, Jay.

0:02:28.0 Jay Feng: Yeah, thanks for having me. I definitely think it is a dance. I would say that for sure.

[laughter]

0:02:35.6 MH: Yeah, it’s like one of those middle-aged dances. It’s very formal and you have to do all the steps the right way and it’s weird.

0:02:45.0 JF: It’s definitely weird. I think it’s also… It’s pretty extreme in terms of level of preparation nowadays, given how much is at stake. And I’d say that it’s like this ritual dance. It’s not like your life is at stake in terms of maybe what you were just referencing with a medieval dance, but more so like your career trajectory is on stake as well. It’s definitely nerve-racking, right?

0:03:09.0 TW: Yeah. Or it can feel like that, for sure. Well, I’m interested just as a starting point, as you progressed through your career as a data scientist, what got you to an intersection of how do we address this problem that you were seeing? And what made you steer yourself towards, “How do I address this more holistically?” Kinda, “What made you more interested in solving this?” I guess, is the question.

0:03:35.9 JF: Yeah. So for me, I was a data scientist for a couple of years and realized that every interview that I had was definitely different from the last interview I would basically have in a loop across different kinds of companies. So, for me, when I was going through the interview loop at Nextdoor, I was realizing exactly how convoluted it was in terms of the different kinds of assignments, the different kind of people that were asking questions, and there wasn’t a lot of structure, I think, for data science interviews and there isn’t a lot of structure right now as well. And so I think for me, it was a natural problem that had to be solved. At the same time, I just enjoyed writing about problems and stuff, and so it combined two interests, I think at the time.

0:04:24.6 MK: So one thing I do wanna delve into is this assignment bit. Are you seeing broadly across… ‘Cause I know at Canva we do a take-home test or you can do it in the interview, if you’re interested. But is that something that is across the board now for data science? ‘Cause to be fair, I feel like there is a lot of variety in interviews, but the bit that I sometimes struggle with internally is the whole take-home assignment bit. Is that stock standard now?

0:04:58.1 JF: Yeah. I think it is for companies that don’t want to spend a lot of time on finalizing a process with different kinds of interviewers and just wanna set up an example of what work you might do on the job. At the same time, there’s kind of this trade-off. This is something that I like to call just in terms of the take-home assignment trade-off is that you’re going to get a lot of candidates that will drop off at this stage because they don’t wanna do the take-home assignment because it just makes the process longer and they don’t get a lot of feedback. At the same time, you also get candidates that are more willing to do the assignment if they enjoy the fact that they get to do the research, they get to do the work, and then not feel like they’re stressed out in some sort of interview format where they’re trying to answer questions off the top of their head. So it does actually leave room for a very particular type of person that’s going to go through your interview process, and so you’re kind of self-selecting at times as well.

0:05:53.8 TW: What always feels like the assignments, and this is the bulk of my exposure to this whole world, I feel like, is just comments that crop up on Reddit in the data science or analytics like subreddit where it does feel like there are people who are way over their head with some particular tool and they’re mid-interview and they’re trying to get help. And I kinda wanna… It feels like if you get an assignment… If it’s like, “I don’t even understand what this assignment is.” I had somebody I know quite well send me the assignment for her organization and I looked at it and I thought, “Yeah, I would not be able to just do this without thinking, but I could get through it if I was going to pursue a role at that company.”

0:06:47.0 TW: But I feel like there are sometimes people who actually get the assignment and do they actually try to fake it? They’re way out of their depth, they go to the internet, they go to Reddit or to friends and say, “Help me do the assignment.”

0:07:00.8 MK: I’ve totally faked it. Not by getting friends to do it.

0:07:05.2 TW: Not in your current role?

0:07:06.0 MK: No. Actually, and also not in my current role, ’cause in my current role, you have to go through your take-home assignment and then they ask you to tweak stuff on the fly, so you’re doing live programming.

0:07:16.4 TW: Well, that’s what I assumed, you’d be able to sniff that out pretty quickly. Like if you’re interviewing for a role and you totally… It’s just gonna bite you in the ass, you’ll feel horribly awkward.

0:07:25.9 MK: Well, in full disclosure, I’m just gonna tell this story because it’s a while ago, and I’m not gonna say which company it was, but everyone will be able to guesstimate. I hadn’t worked in analytics before, and I got asked to do some analysis in Tableau, which I’d never used before. I had no idea what SEM and SEO were, but I was like, I’m just gonna learn. And I was open with them though, that I didn’t have the skillset, but I would try and figure it out. And I didn’t tell them that I spent weeks on it, and I went through Tableau and figured it all out. I somehow got hold, through my former employer’s email. They thought I was a big shot who was buying it for a department, and so they contacted me and were like, “Do you need any help trialling this product?” And I’m like, “Yes, I don’t know how to load data in Tableau, it’s too big.” And they’re like, “Don’t worry, just send it to us, we’ll do it for you.” And so I managed to hack my way through this process, but actually, if I had someone do that for a job from me, I would be okay with that, because the skillset you’re actually showing is like, I’m gonna do whatever it takes to get this job. And if you’re gonna do that in the job, I’m okay with that. But I don’t know, I feel like Jay might just be totally disgusted right now.

0:08:42.0 JF: I do think that the expectations are wildly different. ‘Cause you’re right, someone could spend a few weeks doing assignment and then someone else could literally just put in the allotted recommended time, which is almost always like three hours, and do the assignment. It’s not an equal representation of work.

0:08:58.2 MK: And it never takes three hours.

0:09:01.2 JF: Well, the reason why it never takes three hours is because everyone else is doing it more than three hours. But if everyone did do it in three hours and then it’d be fine, right? Because then they would actually have that hard stop and you could see how much they actually got done. But at the end of the day, the average amount of time is just whatever the average amount of time the candidates decide to take. And if you’re under that, then you don’t have a really good representation of your effort.

0:09:26.7 TW: So where do you start with telling someone to… If you’re coaching someone on prepping for… Will they show up and say, I’ve got this interview coming up in a week, what do I do? Or is it better to even start before that while they’re actually applying for jobs, and part of the interview prep is finding the right jobs to apply for? Where does it even start with this whole topic?

0:09:53.2 JF: Yeah, I agree. I think the interview and the job hunt is just basically melted together into this larger question about career transitions. And there’s no easy questions for any of this, but I would say that on standard, I think most people should spend around two to three months studying, prepping and just thinking about this huge life change that they’re embarking on. Because I actually think it’s important to really, really finalize and figure out what you’re gonna do. And the people that actually just do it more kind of ad hoc, or if they’re maybe interviewing to just see what their market rate is, they can afford that luxury of like, I’m gonna study a week before or I’m just gonna interview a little bit and try to see what’s out there. But in general, I think the right amount of time is around two to three months of just kind of starting looking for jobs, understanding your skillset, understanding exactly what you wanna do next, because it’s such a big question. And everyone kinda treats a little bit too small in scope in terms of like, I got this interview next week, I just wanna pass it. And that’s not really, I think, the right framework to think about it because you’re literally signing up for two to four years of your life on average, on what you’re about to do in the next week.

0:11:13.2 MK: But I feel like there’s such variety in, I guess, whether you wanna say data science or some other term. There is so much variety in the type of jobs and what people mean by that term and the type of work you’re doing. Even just sifting through to figure out like, yes, this company is a good match for me, but yes, this role is a good match for me. It is like a job, right, to try and figure that stuff out. And I love the fact that you’re saying it should be a two to three month process of really figuring out what you wanna be doing and where you wanna be.

0:11:48.2 JF: Yeah, exactly. And I don’t think that all has to happen when you’re unemployed or you’re transitioning in between, I think it’s more of a… A lot of the times, I feel like most people actually just transition within their own job. They start doing a lot of machine learning work or something, and then they realize they don’t like it. Or they’re making Tableau dashboards and they’re like, I’m kind of sick of this, I’m gonna start transitioning on other work. And then they do 20% data analytics work or data science work, and they’re like, “Oh, I kinda like this,” and they increase that amount and then it’s like 50%. And then maybe they look for a new job at that point, or maybe their manager allows them to kinda just completely transition to this new role. But I think the interview is kind of like that stage where it’s so much more officialized, whereas in our day-to-day lives, we probably just kinda shift between this all the time, and yeah, I think it’s just like a natural interaction.

0:12:40.2 TW: So you just described basically my career evolution, which is when I’m talking to people… ‘Cause I will tell, do more of the stuff that you like and are interested in really more, ’cause you wanna make sure. But that even maybe includes learning. And if you do it well, and go above and beyond, opportunities will present themselves. Let me insert the caveat that I am a upper middle class, white cis dude, so I’ve had all the leg up on that front, so easy for me to say, yes, I had all the legs up. I’m like, gotta throw that in. ‘Cause it sounds like you started to say two to three months, and I thought, how often… If somebody is saying, wow, I’ve started to do some of this machine learning stuff and I really like it, but I’m just fumbling along, I wanna find a job where that is the job. Should they be saying, I’m gonna go do a boot camp or a course and try to get to a certain skill point and then go for the job? Or should they say, no, I just wanna say, this is what I’ve done, scrapped together on my own, and then try to find a job where they say, yeah, we’re happy to bring you on and have you learn or… Or is there no magic answer? I feel like cramming for an interview on skills feels wrong.

0:14:08.7 MK: I’ve totally crammed. I feel like everything that you’re like…

[chuckle]

0:14:12.5 TW: Like on what? But what were you cramming?

0:14:14.7 MK: Oh, so I had flashcards because I wanted to memorize all the terminology for stats that I’m really shit at, and like how… What the definitions were and how I would use it in a working example and things like… I’m, but…

0:14:30.9 TW: Were there things that you had used and you just were like, I wanna make sure I remember how I used them? Or were they things that…

0:14:38.9 MK: Mostly yeah, mostly I understood the concept and I might have applied it or not applied it, but it was like… Yeah, I’ve been in some really full-on technical interviews where they are like, “Describe this function.” And I’m like, “Fuck, I haven’t used this function in like four years.” I can broadly tell you what it does, but I can’t tell you the inner workings of it, so yeah, I have flashcards and I do, I memorize. But I mean, Jay could be… Oh my God, you’re gonna be saying everything like Moe is doing all the stuff you shouldn’t do.

0:15:11.1 TW: Moe is the arbiter. I know, he’s probably saying like, no, that’s the… I don’t know what’s…

0:15:14.3 JF: No. You’re partaking in the system that I think is fundamentally flawed or I like… I think if anyone is studying flashcards for anything, you’re just… You’re basically just gamifying yourself, you’re in some sort of system where you need flashcards, but we should never have any system where we need flashcards, just, in my opinion, even for school, or maybe even if there’s some memorization stuff that you need…

0:15:39.3 MK: That’s how I studied.

0:15:40.4 JF: Yeah, yeah, yeah. And maybe that works for some people. It never worked for me. And I think… I think that if someone’s asking you to describe a function, it’s literally, it’s like, “Why do we need flashcards now?” Or “Why do I need to repeat this to you in the interview, we have Google now.”

0:15:55.8 MK: Totally.

0:15:56.8 JF: We have technology now to use it. So if you’re asking me this, then I distrust your interview process even more now, to a degree of which I will excuse myself from the interview. Obviously, not everyone has that opportunity to do so ’cause they need a job or they need to kind of get that leg up but I think in this day and age, no one should be asking questions like that.

[chuckle]

0:16:21.4 MK: We let people use Google in our technical interview ’cause we’re like, that’s what you would do on the job, if you don’t know how to do something, you would go to Stack Overflow and you’d read it. So we encourage people to go to Stack Overflow and be like, “Alright, read up on this function, say what it does, and then if you need help, let us know and we’ll work through an example with you.” In my view, any kind of technical questions or assignment or whatever it is, it should be a two-way street. I walked out of my Canva interview when I’d learned four new things in R, which I didn’t know, and I want everyone to have that experience because that’s what your work is gonna be like. You’re not gonna know the answer to everything, and you’re gonna have to work with your team, and if your team are willing to show you how something works, so like talk you through a function, that’s where I wanna work.

0:17:07.8 TW: Can I just, in defense of the flashcards, I think there’s a degree of whatever makes you more comfortable and feel prepared. If you feel like there’s gonna be a lot of stats questions… I think there is probably a part that whatever… There’s gotta be a degree of confidence, you don’t wanna walk in with bravado like you know everything, but if you’re like, “You know what, I haven’t done that in a while I enjoyed doing it. I wanna kinda flip back through it, I may wind up going to that as an example, and this will make me… This will give me the mental… So like, to me, that is another way to look at it that it’s like you doing what, you are going to hyper-prepare for anything. I know that… Podcast episodes may be excepted. But…

0:17:50.8 MK: But so, like to add context, actually, that particular interview I was doing the flashcards for, everyone who was interviewing me was a white dude in their 40s and 50s, they all had PhDs in Maths or Stats. And I knew it was gonna be stats heavy, and I also have an inferiority complex about my technical skills anyway, so it’s very possibly as much about me and not the process, let’s be honest.

[chuckle]

0:18:19.6 TW: But I think there’s also the part that… I think, Jay you’re… There’s a part of it that if their process is terrible, that’s not necessarily the interviewee’s fault, they may struggle to find good candidates, it’s not the interviewee’s problem to solve a completely broken interview process. And then I think the other, which again, easy to say, not having interviewed in over a decade, to say I want the interview process to turn up if I’m not a good fit. I remember times in my career where I went through an interview and I walked out and I’m like, they’re not gonna offer me the job, I don’t want the job, it’s… I’m not excited about it. They probably don’t think I’m a great fit. And that’s good. So there’s a degree of say… If I’m walking into a bunch of middle aged PhDs like I am who I am, and I assume you don’t fall all over yourself saying I’m terrible, I don’t know anything, but at the same time, it’s the point of the process. You’re trying to find a good fit between the role and the person.

0:19:24.0 MK: But it does affect your confidence, and… So a friend of mine, who is a woman who is absolutely fucking amazing. She’s a woman of color, is going through an interview process right now, and she was like really nervous about it, and was like, “Oh, I’m worried that everyone’s gonna think I’m dumb or not good enough at my job if I don’t get offered a role.” And I’m like, “This is a two-way street. If you don’t get offered a job at the end, it might be because this company and this role is not right for you, not the other way round.” But I don’t think that… Jay, are you… I guess what I wanna understand is, as you’re helping people through this process… I keep coming back to it subtly and not realizing it, a lot of this is confidence too, so like…

0:20:07.9 MK: Is there, How do you prep people for that? ‘Cause you can prep people for answering the star questions and all that sort of stuff, but part of it is also, how do you make people feel confident in their own skills so they can put their best foot forward?

0:20:24.7 JF: Yeah, that’s definitely a tougher one because, going back to the earlier question about, essentially understanding that it’s not a good fit and being okay with it when you walk out. That’s something that I don’t think a lot of people have that understanding of which they always think it’s kind of a one-way street. More offers equals better, and that’s not always the case. And I would say that for a lot of companies, the way that they hire in the interview process, really does reflect on kind of just like what candidate they want, especially, I’ve learned this since I’ve become a person that now hires people.

0:21:01.9 JF: And even if I’m coming into an interview and I’m giving them access to Google and I watch them, how they use Stack Overflow, that’s a reflection of the fact that I want someone that can use Stack Overflow that can use Google to do things. And a lot of companies, they don’t want people that just use Google and Stack Overflow everything. They want you to know it, or they want you to memorize, or they want you to have this down because you’ve done it so much, and I think a lot of software engineering interviews are like that because of the fact that they ask you these very specific, LeetCode-type algorithm problems, and they wanna know that you can just do this in 10 seconds and then take on a higher complexity of a problem.

0:21:45.2 JF: So I think that it really does depend on the company in terms of what level they’re hiring for, what kind of responsibility you’re doing, what kind of worker you are, and that’s why that interview fit thing should be more evaluated, it’s not a lack of confidence but more so, a right match, and this is just not, not something that a lot of early career people know how to basically comprehend, because you go to school and then you go through the system where everyone is on a bell curve, and it’s pretty obvious that you did well or you didn’t do well. Or that you’ve got good grades, or you didn’t get good grades, and I think that it’s not really something that obviously applies into the real world as well and, yeah, it’s definitely something that is hard to kind of tell people at scale, I would say. This is all kind of through self-learning, and I’m not really sure if you can kind of listen to a podcast or read a blog post and kind of accept this fact, about yourself and about your career but definitely something that, I myself have kind of learned after going through hundreds of interviews throughout my career.

0:22:49.8 MK: How do you help people prep for that two-way process of getting candidates to understand that it’s as much about them interviewing the company as it is about the company interviewing them?

0:23:02.3 JF: Yeah, that, that’s really tough. And we do do coaching sessions on Interview Query, which kind of accentuates this fact, but for the large part, our platform is built, so that anyone who comes in can be successful on any given interview. And specifically the technical portion, and I would say our company is meant to kind of help you prepare for a very specific portion of the interview, but a lot of our success is dependent upon the fact that you yourself can accept that you’re not gonna get a lot of offers potentially, or it’s not gonna be a right fit with a lot of companies as well. And so for these one-on-one coaching sessions, I think that’s where we try to explain the nuances of this, because anything you research online is gonna be so context-specific and kind of taken out of the general context that you’re gonna be learning about. Most people on any kind of form is gonna talk about, how it’s better to be confident, or to practice leadership questions or to pass all your interviews to get more offers, but no one’s really gonna walk into the nuances of who you are because they won’t really have that conversation just with you.

0:24:11.0 JF: You might be able to find someone who’s very similar to your background, if you’re just a new grad or you’re transitioning from another field, by the end of the day, it kinda really depends on who you are as a person, and that’s where you can really get that only with just one-on-one conversations and coaching and stuff.

0:24:26.5 MH: So one question I have that take us a little bit different direction, but I think it’s something that a lot of people ask about is what level of education? Especially in data science roles and I know you just did a YouTube Short about this, I’m kind of setting you up, but…

[laughter]

0:24:45.2 MH: It’s interesting because a lot of people ask that, they’ll get out of their undergraduate degree, they maybe do a couple of years as an analyst, or a junior data science role. But then it’s sort of, “Do I need to go get a Masters? Do I need a PhD? Or should I stay in school and get a PhD? And is that the way I need to go? What track do I need to be on? If I wanna go into a in career data science.” It’s sought of, what are you finding out there? Sort of what are the, and I think there might even be levels to it as well, in terms of what kinds of data science roles as well? But what have you kinda learned about that, What advice could you give to people who are kind of, “What we have, what should I do if I wanna pursue that?”

0:25:20.7 JF: Yeah, the education question is probably the most common one and obviously really difficult to answer for every person. I’d say like… The most general answer I can give is that, the biggest misconceptions are, is that most people think that they need a Masters, but most of the time, it’s just the fact that if you can get into a Masters, then you’re, they want people that can get into a Masters and go through a Masters, they don’t need you to go through and get the Masters.

0:25:50.2 MH: So just go apply, get approved, and then stick it on your resume. [laughter]

0:25:54.6 JF: Exactly. Right.

0:25:54.7 MH: That’s awesome.

0:25:57.9 JF: They just, they’re using education as a benchmark for what can you do? It’s just like anything else, how much experience you have? And it’s just too easy to use a Masters degree as a way to kind of eliminate a certain portion of the field, because right now, we have an over-abundance of data science candidates, almost all ways, in any sort of position and you really do need to filter it down and you’re gonna get false positives and false negatives there, everywhere. So in general, people that like structure and people that like schools should absolutely get a Masters, if that’s the only way they can effectively learn. The people that can self-learn and do projects and be motivated no matter what, should just do that route, instead of going and paying for a Masters, but at the end of the day, it’s more so, how much time can you dedicate to focusing on Data Science and a Masters degree will definitely force you to focus on data science for about 8 hours a day, I’m guessing.

0:26:54.1 JF: Yeah, but so will another job at the same time, right? And so will like you just sitting down and doing it as practice and just preparing for interviews. So a lot of it is just kind of like time spent just preparing. Now, I think the only caveat here is that there are certain like research science jobs that require like, you make some sort of breakthrough. Like I don’t think Google X or Google Research is going to hire a Bachelors degree, you know, a fresh college graduate unless if that’s how it is, so.

0:27:24.1 TW: Even with the Python, a six-week Python boot camp?

[laughter]

0:27:28.0 JF: Yeah, maybe with a six-week Python hoot camp, you might have a better…

0:27:32.5 TW: Yeah. Shoe in.

0:27:33.1 MK: Out of curiosity, do you see any difference like on your platform about I guess the self-taught six-week boot camp-type candidates or people who do have degrees? Like do you notice any difference coming through in your experience or is it just like it really depends on the person, and how motivated they are?

0:27:57.3 JF: So for outcomes like who gets the job or just like what our audience is kind of you mean there?

0:28:02.0 MK: Yeah, but also just like there’s a technical, like it’s all about technical upskilling, right. And I just like, I obviously have my own personal experiences, but I’m curious to see what you’re exposed to.

0:28:13.1 JF: Yeah, so in terms of technical upskilling, I would say that, most of the time we see people that come in, and they want a little bit more structure towards what they need to do to basically upskill themselves, right? And so like for us, like what we do, it’s pretty easy, but all we do is we take a problem, we basically make it into like a bite-sized kind of one-hour time block of a problem that you should solve. And then we hand it to you, and we tell you that this is generally an easy problem, or generally this is a medium problem, over like 60% of the people that have been able to solve this problem are gonna be career ready. And so we’re kind of just standardizing the field so that you understand like, this is what is required probably to do this job well or like this is what, this company will ask on the interview because that’s what they do on the job normally. And I think that’s a great way in my opinion of just like really taking like this complex field and just kind of narrowing it down into like a single kind of one-hour step, so that you can just kind of, like, you can like, pretend to, like, do the job for like one hour, right?

0:29:24.3 JF: And I think that’s what a lot of interviews really are. It’s just trying to understand if you can, like, do the job within this limited time window. But when you’re being exposed to these kinds of problems, then you’re basically just practicing the way that you can think about data science, basically in like a timeframe. And I think that’s really valuable because otherwise it’s really difficult to just go out on your own and try to understand and replicate that environment. And I think a lot of people initially would go on Kaggle, and they’d sell a bunch of Kaggle exercises and look at other people’s code and be like, “Okay, I feel pretty confident for like the job interview now.” And then you go on the job interview and then bomb it because you realize that it’s different than doing a Kaggle exercise. And I think that’s just like the…

0:30:08.1 TW: But I know who’s going to die on the Titanic.

[laughter]

0:30:10.8 TW: We’re selling shoes.

0:30:14.9 JF: Yeah, it’s like you’re telling me, Google doesn’t solve the Titanic problem.

0:30:19.0 MH: Yeah, that’s right.

0:30:20.6 TW: Well, I mean, but that is for the big tech I assume, whether it’s Nextdoor or Google or Amazon or Meta, that it feels like there’s like people know, people know what the interview process is like, and I think a good recruiter regardless, even if it’s not that large of a company, the recruiter it’s kind of their job to prep you a little bit with like this is the interview, but like where? I mean, I know Moe tends to, you’re probably going to give the… I totally stalk the crap out of it. Like…

0:31:00.8 MK: Oh yeah. I stalk.

0:31:02.4 TW: What’s the extent to, I’m going to go interview with this company that I have not heard of before I saw the job opportunity, I’ve researched the company, seems like they’d be kind of interesting? How deep should you go on trying to find people inside or find what the interview is really like or what they’re really looking for like, the sort of stealth research of what is it they’re looking for? How are they going to approach it? Not necessarily to the point of what coding question are they going to ask me? Like is that a worthwhile investment to try to get intel through any means necessary.

0:31:45.1 JF: Absolutely, right, imagine you had to like the next person you were dating, right? Like, how much do you stalk the person that you’re going to go on a first date or like, maybe the fifth date with before you jump in? Maybe, maybe it’s nothing, or maybe it’s a lot, but like…

0:32:02.7 TW: Oh, I usually ask my wife to talk to her a bit and see what she thinks.

0:32:06.8 MK: I feel like this, you should stalk more for this than for a date, personally.

0:32:12.3 JF: Yeah, I would say so too, because, you know, you might go on many different dates and then that’s a little bit easier match because it’s only one person. But I mean, think about this, right, it’s eight hours a day for however long you want to stay at your next job, that’s a pretty big time commitment. And I feel like a lot of people’s unhappiness is also attributed to just their career and maybe just not getting the job that they really wanted. Now, obviously, there are some other conditions here around like we said before about getting income and putting food on the table and stuff like that. But like I think that anyone should use every kind of advantage that they can to really understand if this is like the job they want, if this is something that they can get an advantage towards landing the offer just because, you know, why wouldn’t you when it has such a big impact onto the rest of, a few years of your life essentially?

0:33:07.8 MH: Yeah, it’s interesting like, I think, you know, because our industry is in such demand, I think sometimes I’ve found that people don’t prepare as much if they don’t perceive the role as being important to them, like in when I’ve interviewed people. And maybe it’s ’cause I was at a small company and not like Amazon or something. But it’s kind of like, yeah, it’s… I’m not sure. And there’s a lot of turnover in our industry as well. So how can people look at that and figure out like, “Okay, is this the kind of company I wanted to work at and make a better choice about how do I stay there longer than a year, or 18 months in the first place?” ‘Cause like…

0:33:51.6 JF: Yeah. And I think a lot of that turnover is really dependent upon a lot of people not doing that research initially, right? Or just going through the interview process and not asking the right questions, or maybe even being misled in the interview as well. But from what I understand, I think most of the turnover and attrition, and kind of not feeling that right fit, is from just the company not understanding that the individual wanted some role, and then the company delivering a different kind of role. Or there’s some just general mismatch going on. And I think that it’s pretty common in data science, because there’s not that much structure. If you think about every other kind of role, like a software engineer generally understands what they’re gonna do once they jump onto the job. And I think it’s more than about kind of culture fit and like what kind of little small things, like co-reviews, or kind of team bonding things that kind of happen. But for the data science role, there’s definitely just so much more kind of variety in terms of like who… What team they’re gonna work on? Who are they gonna interface with on a day-to-day basis? Are their projects gonna be like super long, and kind of very grand in scale? Or are they gonna be kind of pulling reports, or doing really quick analyses, right?

0:35:07.2 MK: When they say data science, do they mean actually like data analytics, or do they mean like machine learning engineer, ’cause there is this huge range.

0:35:16.2 JF: Exactly, right? And then if the interviewer doesn’t really know to test for those specific skills for that role, or maybe the interviewer is a machine learning engineer, and then they’re hiring for a data analyst, but then they only know how to interview for machine learning roles, then what are they going to give that person? And then who’s the person that’s gonna actually accept the offer, and then suddenly understand that they didn’t want that role? So I think it’s, at the end of the day, it’s kind of a… It’s all about kind of getting that fit, right? And I think because it’s a multi-sided marketplace, you can understand why hiring is this huge business, because it just continuously churns out attrition and bad fits, and then also maybe some good fits that could 10x a company. But at the end of the day it’s just like a huge problem because there’s so many different kind of parties involved, and with so many different kinds of incentives at play.

0:36:05.9 MK: Out of curiosity, one thing that’s on my mind, like a lot of what we’re discussing to me does seem to be at the more junior to mid-level data science roles. Are people still preparing for technical interviews once they get to more senior levels? Or is like, does that kind of requirement drop away? Again, this is not something I know if you know, but it’s just something that was bubbling away in my head.

0:36:32.5 JF: Yeah. I would say I’ve talked to a few people, and I think that they still do a little bit of technical interviews. I think there’s probably at least one round of technical interviewing, whether it’s actually doing SQL or like coding stuff, or it’s actually specifically around kind of like a case study, or kind of a larger strategical question about how you move the data science org and push it forward, right? But I think there is this understanding that you have this backbone of technical knowledge, and now you kind of are moved into this role where you’re kind of like more into the operations of how you get throughput and kind of effort, and deliverability out of your actual data science team. And there’s definitely a lot more kind of behavioral interviews at that stage and kind of like leadership expectations.

0:37:24.3 JF: But I think it’s very much like because this is a world where the general end user, and the kind of audience is also so niche. There’s probably, I don’t even know, there’s probably just thousands of data science managers and leaders. I don’t think there’s hundreds of thousands even. It’s just such a small kind of area in which you don’t really have a lot of information going on there, right? And it’s definitely gonna be just be super dependent upon the company as well. And so I think data science in general, definitely probably like what much more kind of structural knowledge. But as we kind of get up there into leadership, it’s definitely… There’s just not as much kind of information around like, “What really matters for leading a data science team and actually doing a good job, doing it as well?”

0:38:11.0 TW: Yeah, seems to get thinner and thinner. That actually brings up sort of another really persistent question. And I know I think I get a lot, which is, individual contributor versus management route in yours… As your career grows, or progresses, and how do you kind of address that, or determine what you should do, right? So it… I think a lot of people struggle with that. I don’t know, would you, or coaching people, or talking to people, how much is that coming up, and how do you guide people through kind of the decision process there?

0:38:42.5 JF: Yeah. That’s pretty tough. I think most people after they have five to seven years of experience, kind of understand generally which way that they wanna go into the management track, or the individual contributor track. Or even if they wanna stay in data science and analytics, to be honest. ‘Cause I know a lot of people kind of transition out into different roles just to test things out. But I definitely think that the data, like the individual contributor route is probably… It just gets more and more kind of technical and more and more so you work on kind of specs, and you’re focusing a lot more on just deliverability of general data science models and projects. While on the manager route, you’re definitely doing a lot more of strategic decision making, and trying to coach your data science teams towards delivering more insights, and more analyses and holding more sway over the general company as well.

0:39:44.2 JF: So it definitely… I don’t know, I think people should have an idea or understanding, ’cause they’ve probably seen a little bit of all of it, by the time they’re at like five to seven plus years of experience, when you start diverting into those routes. But it’s a tough career question, for sure.

0:40:02.2 TW: And not to tip our hands, Michael, but I think we should, maybe we have an upcoming episode where we…

0:40:08.2 MH: We dive into that in a little more detail.

[laughter]

0:40:11.0 TW: Dive fully into that specific topic in more detail, so…

0:40:13.9 MH: Perfect.

0:40:14.5 TW: Or maybe you’ve popped over to the wrong show prep document there.

[laughter]

0:40:18.5 JF: No, I think it’s worth discussing, because I get that question persistently, and I know I struggled with it myself too growing up.

0:40:29.5 TW: And I wonder is that transition, is that easier to interview externally for, “I’m going from I’ve been an individual contributor and I want to make the leap to a management or a leadership role?” Or is that really an easier transition to say, “Well, I moved into a full-on Supervisory Management or at least pseudo-guidance role internally, like I took on a mentoring role.” It feels like that is from an interviewing… If they’re looking for somebody to lead people and you’ve only been in a pure individual contributor, then probably your responses need to say, “Well, technically, I’ve been an individual contributor; however, we had two interns who I basically oversaw for that summer.” Or that would be a pretty big leap for an organization to make that you’re coming in from externally, and we’re gonna drop you into a whole range of skillsets, that if you can’t articulate why you have them, “I just wanna get into management,” that feels like it could be tough.

0:41:39.0 JF: Yeah, I think so too, and I think historically it’s always been that you had to either join a startup that was growing really fast so that you as like an early adopter…

0:41:50.9 TW: Just by default.

0:41:51.7 JF: Or early employee. Yeah, by default you just become a manager because they need you, or you had to then somehow find a way to manage interns or mentor employees and then grow into that role, or switch and talk about that experience. But at the same time, I don’t think they like external hires that haven’t had that management experience before, because it is kind of a bigger risk, so…

0:42:17.3 MK: I actually find recruiting for managers, like data managers is one of the hardest things to do. There actually doesn’t seem to be enough of them. This is in my experience, or like…

0:42:29.0 TW: There’s an abundance of people who want to be them, I think.

0:42:31.4 MH: Yeah. That’s the thing.

0:42:32.4 MK: Really? Because I have probably three…

0:42:35.4 TW: No, no, no, not necessarily ready or able or capable.

0:42:39.6 MK: Oh.

0:42:39.7 MH: Or should ever be, yeah.

0:42:39.9 MK: No, but we…

0:42:40.0 TW: Or should ever be, like that’s… Yeah.

0:42:42.8 MK: I would say probably three people that are like, “I wanna be on the coaching track, like the Manager track.” And then they give it a red hot crack and then they’re like, “Actually no, I don’t wanna do this, I wanna go back to being a Senior IC.” And I think part of that I do also think is creating a pathway for people to be senior and be a technical lead without necessarily managing people, which not every company has, but I feel like us hiring managers is really hard. And maybe it is, like you said, it’s like you do expect people to already have that experience before they interview with you, so you’ve got a more limited pool versus people that want to do it that may not have had the opportunity yet, but…

0:43:25.7 JF: Yeah, I think sometimes it’s like the ICs test out to be managers, and then they’re like, “Wow, this job is just meetings and I don’t do anything.”

0:43:34.3 MK: An admin. Yeah.

[laughter]

0:43:34.5 JF: Yeah, in an admin, exactly, like hiring people and maybe firing people, and stuff like that, and it not really feels as much output you’re actually contributing towards the company versus just doing the hard work. But I think it still is kind of work and it’s still actually really important, it’s just that you just… It’s a fundamental shift, in terms of what you perceive as work and actual what value you’re contributing to the company. So I definitely could see how that gets really, really exhausting, or maybe it’s just like a weird shift in…

0:44:09.2 TW: Well, there are people who are great at it. Michael gets energized by helping people grow their careers and be successful.

0:44:18.2 MH: Yeah. But if I had built up any data science skills, maybe I would’ve preferred that, so. [laughter] No, it is an interesting question. But we do have to start to wrap up, Jay, you’ve aced the interview. Thank you. No, I’m just kidding. It’s not really an interview. You’re hired. You’re good. No, thank you so much. So one thing we love to do is go around and share our last call, something we’ve found recently or want to… Think might be of interest to our audience. So Jay, you’re our guest, do you have a last call you wanna share?

0:44:54.7 JF: Yes. On my last call I would say that I’ve found a lot of really, really interesting new startups around generative AI in the past month, as I’m sure everyone else has been kind of like playing around with the tools. So for me, I would recommend everyone to just kind of play around with these prompt engineering-like tools, like Playground, AI.com, another one is Alexika. I think these are really, really interesting. I think everyone will get a kick out of it, who’s in analytics and data science.

0:45:28.4 MH: Nice.

0:45:29.7 TW: Awesome.

0:45:29.9 MH: I love those. They’re interesting and strange all at the same time. And very hard to understand like, “How is a computer doing this?”

0:45:40.4 TW: It’s funny, I had a potential last call that’s one of them for some future episode that somebody pointed me to. I need to dig into it more before I use it, now it’s gonna be out for a while.

0:45:50.7 MH: Oh, see. The way to tease it there, Tim. Well, why don’t you share what your real last call is then, Tim?

0:45:57.2 TW: Well, my real last call, this is oddly… It’s gonna be very, very short, and it is not really analytics.

0:46:03.7 MK: That’s weird. The short bit.

0:46:06.0 TW: But I’ve found… Well, yeah, well, and I’m not done yet, so we’ll see. But so it’s a podcast called Pod Save the World, which totally is gonna reveal my progressive leanings, which…

0:46:19.7 JF: I heard that. I listened to that podcast…

0:46:20.0 TW: Yeah.

0:46:20.0 JF: Recently, too. It’s random.

[laughter]

0:46:24.1 TW: I, wow. I’m a little bit of a… The whole crooked family, but one of the co-hosts of that is Ben Rhodes, former deputy National Security Advisor. Nothing to do with analytics. However, on a recent episode, I’m gonna play an audio clip from Ben Rhodes that I just, well, I’m just gonna play it and I’ll get your reactions. So I’m gonna play it now.

0:46:49.7 MK: I love that you said it’s gonna be short and now you’re playing us a clip. I’m all for it but just…

0:46:56.6 TW: It’s one second.

0:46:57.5 MK: Okay.

0:46:58.7 TW: It’s one second.

0:46:58.9 Speaker 6: Do your fucking job.

[laughter]

0:47:07.4 TW: So I was like, that’s my guy.

0:47:11.0 MK: Did your brain explode when you heard that?

0:47:13.0 TW: Oh, I got a bit. I was running, I was this guy 5:15 in the morning with this big shit eating grin on my face. I’m like, you go. He was obviously…

0:47:22.4 MK: So for poor Jay, who has no idea what the fuck is going on. Tim says this particular quote very, very, very frequently. It’s like his catch cry.

0:47:32.1 JF: Yes.

0:47:32.5 MK: And the fact that someone else shares his thoughts will…

0:47:35.7 TW: This is, it’s why I’m not in management.

0:47:37.3 JF: That’s right.

0:47:37.8 TW: Because my, I do not have the bedside manner for…

0:47:40.9 MH: Yeah, yeah, beautiful.

0:47:41.4 TW: Data leadership role.

0:47:42.8 MH: That’s probably gonna be a copyright strike against us, but we’ll allow it. Anyway. Okay. Moe, what’s your last call?

0:47:49.9 MK: I just finished reading Daniel Pink’s book ‘When,’ and we know that I’m a big Daniel Pink fan, but this one was just really interesting because there were so many like how-to books and all that sort of stuff. But this is really about using the best time of day for different tasks, which is not something I’ve really thought a lot about. But we all know that I like struggle with focus and all that sort of stuff. But one of the like, insights that probably resonated with me the best was I often avoid stuff in the afternoon because that’s my down trough, which is natural for most people with their circadian rhythm. But you’re actually more creative during that period. So like, if you’re gonna do like a team brainstorm or something and you want people to think outside the box, actually scheduling it in the afternoon’s a really good idea. Whereas I think before I was like, “Oh, you wanna do it in the morning when everyone’s minds are fresh.” But the truth is, people are very good at like process and analytical tasks first thing in the morning. So it was just one of those like Aha moments that I will probably start adopting like into my scheduling moving forward. So you might not have to read the book now, because I’ve given you…

0:49:00.7 MH: Does that cross over? What’s the sleep book? Is it the… What’s the sleep book, did…

0:49:04.7 MK: Why We Sleep? No. Why Sleep Matters? No, shit. Matthew Walker…

0:49:08.4 MH: Why Sleep Matters?

0:49:09.5 MK: Like the book that I’m obsessed with that I can’t remember the title of.

0:49:14.5 JF: Why We Sleep…

0:49:15.7 MK: Why We Sleep. That’s it.

0:49:18.0 TW: Yeah. Why We Sleep.

0:49:19.0 MH: Why We Sleep. There you go.

0:49:19.0 TW: Okay.

0:49:19.0 MK: Yes.

0:49:19.0 TW: You had it right.

0:49:19.4 MH: I just, as you’re talking about the, when you hit energy and other peaks.

0:49:23.8 MK: Yes.

0:49:23.8 MH: And what’s going on.

0:49:24.7 MK: It is definitely something I think a lot about, about how to like, use the day to my advantage. Not that I like actually implement all of this stuff, but I think about it, which is at least the first step to change. Right.

[laughter]

0:49:37.6 MH: That’s right. As long as you’re thinking about it.

0:49:39.5 TW: So Michael, have you finished the have you finished the your book?

0:49:42.6 MK: Finish?

0:49:42.6 MK: Finished, yeah.

0:49:45.6 MH: Absolutely not.

0:49:45.6 TW: Okay, that’s good.

0:49:46.1 MH: No. I probably never will.

0:49:47.7 TW: What’s your last call?

0:49:48.5 MH: Well, as it happens, I ran across this a little while back Post Band came out with a state of the API report, which I actually found kind of interesting because in data and analytics, we use APIs all the time and we’re kind of working with them. This is sort of more of like a broader view of it, but I had some very interesting information in it, like what roles are working with APIs the most with different companies? Where do we see API growth and trending and those kinds of things. So it’s an interesting report. So we’ll put a link in the show notes if you want to check that out. You can. All right. I know you’ve been listening and if you’re anything like rest of us, you’ve struggled with these interview questions, you’ve struggled with sort of figuring out how to guide your career and where to take it. And we wanna hear from you, so please reach out. And the best way to do that is probably through the Measure Slack Group or Twitter or on our LinkedIn page. So we’d love to hear from you. Jay, are you active on the social medias at all?

0:50:48.9 JF: I am active on the social medias. You can find me on LinkedIn, Twitter, I will post stuff from time to time about data science. So yeah, please check out the show notes or wherever this is going.

[laughter]

0:51:04.6 MH: Awesome. Yeah, and as I mentioned, like you have a YouTube channel and things like that so people could check out YouTube as well, which is pretty cool. Some nice information there. So, great. It’s awesome to have you. Thank you again, Jay, for coming on. I think this is such a relevant topic and for the reasons you described, so really appreciate it.

0:51:22.7 JF: Yeah, thanks again for having me. This was fun. See you around.

0:51:25.5 MH: Yeah. And of course no show would be complete without a huge thank you to our producer Josh Crowhurst. Thank you Josh, for all that you do. It’s a pleasure working with you and by working with you, I mean, whatever you and Tim are doing behind the scenes. All right. I know that interviewing can be tough and it’s hard to know how to do everything exactly correct. And it’s stressful to face potential rejection even if you don’t really even want the job in the first place. But I know I can speak with certainty that my two co-host, Tim and Moe agree with me no matter what job you’re going for, remember, keep analyzing.

0:52:07.9 Announcer: Thanks for listening. Let’s keep the conversation going with your comments, suggestions and questions on Twitter at @analytics hour, on the at web@analyticshour.io, our LinkedIn group and the Measure chat Slack group. Music for the podcast by Josh Crowhurst.

0:52:26.0 Charles Barkley: So smart guys want to fit in. So they made up a term called analytics, Analytics don’t work.

0:52:32.6 Tom Hammerschmidt: Analytics. Oh my God, what the fuck does that even mean?

[laughter]

0:52:42.9 MH: We’re getting close to your annual performance review, so I can tell you’re kinda kicking it into…

0:52:46.8 TW: Well, this is that this podcast couldn’t come at a better time ’cause I’m definitely gonna to brush off my interview chops, if that’s the case. Alright.

[laughter]

0:52:55.2 MH: Do you have any tips for you know, how to have some of that hosts… Podcast host with fast decaying skills looking for high end… The pretty crisp, crisp… Yeah.

0:53:15.7 TW: Yeah, Mike, I don’t know. I feel like we’ve worked a little too professional so far, so, but…

0:53:20.3 MH: Well, you know.

0:53:22.1 MK: As always, talk to James Todd.

0:53:23.4 MH: As you said Tim, performance reviews are coming up, so.

0:53:26.5 TW: There are anomalies even in a…

0:53:29.9 MH: All right, let’s, it’s the, what we call the Jay effect. Jay is having this umbrella halo effect, attribution. All right.

0:53:41.6 TW: Is it the lighting?

0:53:43.5 MH: Yeah. That must be, yeah.

[laughter]

0:53:46.7 MH: Rock flag and you’re hired!

Leave a Reply



This site uses Akismet to reduce spam. Learn how your comment data is processed.

Have an Idea for an Upcoming Episode?

Recent Episodes

#257: Analyst Use Cases for Generative AI

#257: Analyst Use Cases for Generative AI

https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_257_-_Analytics_Use_Cases_for_Generative_AI.mp3Podcast: Download | EmbedSubscribe: RSSTweetShareShareEmail0 Shares