An Interviewing Retrospective

11 minute read


I’ve just recently accepted a new role as a data scientist at Zapier starting May 2022. This concludes a sequence of interviews on, what felt like, every other day. There’s a lot I learned in this experience, and I’d like to share some of that with you. I’m going to omit names of companies where needed, but I’ve tweeted enough for y’all to know who was in the mix for this role. At the end, I hope you learn something about yourself, your career, and interviewing.

About Yourself

A network is invaluable. That sounds kind of trite; how many times have you heard “you need to network”? But, if you’re like me you may have misunderstood what networking is and how to do it. I used to think that networking was transactional; that I was supposed to feign interest in people and problems in order to get a job. I felt like networking happened in a 2 hour time span. It left me feeling cheap and drained so I didn’t do it much, or at least didn’t do it in the form I was told to (coffee and business cards).

But networking doesn’t have to be a room full of people and 2 drink tickets. Of the 10-12 companies at which I interviewed, the vast majority reached out to me on twitter and said “Hey, heard you’re looking for work…”. That’s how networking is supposed to go. On twitter, I attract and engage with like minded people genuinely. Consequently, I’ve developed a sort of “brand” for myself and rapport with people. “That’s Demetri, he is passionate about doing stats right” is my brand (I think) plus maybe some other stuff about being a good guy. Point is, the genuine interaction and consistency in keeping those relationships alive vis a vis stats memes is what networking is supposed to be like. I’m not saying you need to blow up on #statstwitter, but I do think you need to be patient and genuine to get the most out of networking.

I can work for Google. This is not generalizable, but HOLY SHIT I CAN WORK AT GOOGLE (some things got in the way of me actually working there but I did get the OK to be hired). I wrote a blog post in 2019 about being really scared to apply for Google Summer of Code, and now look at me! Anyway, that’s done a lot for my self worth and has also helped me realize what I really value in a job. Spoiler: it isn’t complexity anymore. If I were to exploit a single aspect of work, it would be complexity because I like hard problems. But as always, a bit regularization has helped me avoid extreme conclusions and will (I suspect) result in better outcomes. Is Zapier the same as google with respect to experimentation? I’m not sure they are, but they are experimenting and they offer additional flexibility that Google did not (as of writing this blog post) and I anticipate I’ll be happier because I’ve regularized a little bit. That isn’t to say I will always prefer this trade off, but for now its what I want.

About My Career

Data Science is still poorly defined which is weird because its nigh a decade after Harvard Business Review called it the sexiest job of whatever. In any case, even conditioning on things like title, seniority, industry, and desired experience leaves a lot of unexplained variation in job responsibility. I’ve found that for very mature companies (e.g. Google, Yelp, Twitter), data science means mostly statistics and experimentation. ML was mentioned here and there, but it was mostly statistics as a means of optimizing product through AB testing or similar. For medium sized companies (e.g., data science meant more machine learning than anything. AB testing was handled by “data analysts” (I’m not sure what the difference is there). I didn’t target many startups because that just wasn’t what I wanted, but from what I could tell data science meant a lot of SQL and reporting to them. In finance, data science is indistinguishable from wanting magic. If you have time, energy, and are able to, I would recommend applying to companies of various sizes to see this difference first hand. Google barely asked me about ML, but grilled me on OLS. Voices didn’t care much for my careful statistical analysis and lingered a little too long on variable selection before I said something like “Its a demonstrably bad idea and I would never do it”. Ignore people who collapse data science onto a single dimension, saying you don’t need stats or whatever. This implicitly closes doors you might not have known to be open. Know what you want, go after it, and ask questions of the interviewers to make sure you and the company are aligned.

Also think about what you’re up to learn. I’m not interested in algorithmic trading, and so if my job required a lot of independent learning on the topic, I would hate it. I would hate it a lot less if I had a team member to guide me through stuff however.

Salary doesn’t hit like it used to. Economists can speak to this more thoroughly than I can by appealing to decreasing marginal utility and opportunity cost, but I’m here to tell you that you should take this fairly seriously. The folklore is that switching jobs is the easiest way to get a raise, and from where I stand that seems to be true. Its easy to get starry eyed when some places offer a 1.5x-2.5x increase in salary, but those places can also end up not being right for you. That’s how I landed at the bank initially. I had an offer to do AB testing at a medium sized company for not a ton of money and opted for the bank because the number was more than I had ever see in my life. I told myself I’d never do that again, but its hard to turn down more money.

Try not to put more weight on salary than is needed. This is going to take a lot of introspection to understand how you value money and to what point you’re going to be comfortable. I’ve taken a lot of time to parse this out, and that made turning down one offer for 20% more base salary as compared to Zapier’s offer a little easier (though not completely and sometimes I groan about it). Think about how sustainable the role is going to be, as in how long do you think you’re going to be able to put up with stuff and be engaged, how much you’re going to work and learn, possible career trajectories, hell even think about what kinds of bullets you want on your resume when you eventually leave that role.

Speaking of money, never settle for less. I was panicking up until March of this year because soon I would have to go into the office. I started to do the math and figured if I moved to Toronto, it would be a 15K pay cut (due to rent increases, and other stuff). So I figured, why not just take the pay cut upfront and move to a remote role? No, dumb idea. Always shoot to make more, because when switching jobs you’re in the best position to do so.

About Interviewing

You need to prepare, unless you don’t. Many of the interviews I had were very stats focused, and I like to think I know stats very well. Things would have been a lot different if I had to grind out leetcode type questions (and interestingly, Yelp was the only one to give me a SWE type whiteboarding interview). I’ve also discovered that I need to understand how to answer “how do you manage stakeholder expectations” because that came up in near every interview. I’m still not sure how to answer it.

Take it slow. I was in a big rush to find a job because the bank was going back to the office of April 2022 and I did not want to join them (for a variety of reasons I will not go into). This meant that there were some weeks I had interviews every day, sometimes twice a day. That is not sustainable, and can put you in the (sometimes good, sometimes stressful) position where you have multiple offers at the same time. One interview a day is manageable when they are an hour long, but several companies put me in those 5 hour long interviews, meaning I might have to spend 5 hours talking about stats and work. Timezones made this even worse, because I would sometimes wake up way too early or stay online way too late. The best companies allowed me to do 2 hours of the 5 across a bunch of days, and I would highly recommend that even if you’re just interviewing at one place. You want to bring your best, and taking time to reflect and be deliberate is a good idea.

Some people don’t know how to interview. Its OK to take a small break. One particular company had an interviewer who, from where I sat, was dead set on tanking me. Could it have been a test to see how I acted under pressure? Maybe, but I just think they didn’t know how to ask the right questions or coax out an answer from a clearly nervous and tired candidate. Its ok to be rattled in an interview, and when I am rattled (like I was for this particular company), I excused myself to get a glass of water. Take 30 seconds, mute yourself if its a zoom interview, compose yourself and breathe. Take more than 30 seconds if needed, if you’re going to tank yourself it will be because you panicked and not because you vanished for a hot minute. Come back composed, talk slowly, and try to salvage things. Composure is key.

Be patient with the offers and just communicate. Negotiating is some weird game theoretician’s wet dream. How much information do you keep/let out in order to always keep the upper hand? That view of negotiating frames the company you may work for as an adversary, which kinda seems weird (I mean, they shouldn’t be your best friend but they also shouldn’t be your opponent). I was just honest with everyone this time around and it worked pretty well. “I wanna work remote, this is the salary range I’m expecting, here are the companies I’m interviewing for right now, do you think we can check all those boxes?”. I think I did this because I reached a point where additional interviews were just burdens rather than genuine interest, so I was really picky with who got to have 2-5 hours of my time in a given week. It worked out really well in my opinion, because I ended up having a bunch of really good offers which matched (and actually exceeded in multiple ways) what I was looking for. And in the cases where we couldn’t come to a complete agreement, most companies tried to compensate with signing bonuses and other perks. Just be honest with recruiters/HR people. All you can hope is that they do the same, and it prevents you from over exerting your very limited mental energy.