I've read a lot of interview reports on the forums and gotten a lot out of them, so let me quickly share mine. The position I interviewed for was Quantitative Analyst. HR just asked about my visa status and then scheduled an HM phone interview.
The HM interview was on Zoom. At first I thought there'd be some technical or resume questions, but it ended up being basically all BQ. A lot of it was about my past experience — I'd done end-to-end modeling for many years, and that seemed to matter a lot to them. We talked a lot about how to work and communicate with different teams, how to handle relationships with people, how to evaluate models, and so on. It felt like a really good conversation, better than any HM interview I'd had before, and I felt like I had a good shot at the next round. At the end the HM said they'd get in touch with HR to set up a Power Day for me. I was really excited, and afterward I sent the HM a thank-you note.
The next evening after the HM interview I got an email from HR saying I'd made it to Power Day, with a few time options to choose from.
The following Monday, HR sent an email confirming Power Day and also gave me a lot of prep materials. Afterward I called HR and carefully confirmed what each round would cover — I asked a lot of questions, and it was really helpful!
Power Day was a full day from morning to evening, with breaks in between. My order was: Technical -> Case -> Code -> Role Play -> Quant Case.
Technical was with someone senior. There weren't really any specific technical questions — it was all BQ, the usual standard ones you see on the forums, but with follow-ups tied to my own experience. I felt like they really cared about teamwork and presentation/communication skills — how you handle conflict with different kinds of people, that sort of thing. They also had me name the three things I thought were most important when evaluating models. The whole thing basically filled up an hour.
Case was the amusement park question — it's a well-known one from the forums. First he asked how I'd make money if I had an amusement park, and said I could think of it as something like Disney. The second question was calculating profit. The third was calculating three different types of passes — one pass had a certain number of visitors per year, he gave the total visitor count for the year and you subtract to get the rest. I got stuck for a moment in the middle and the interviewer gave me a hint — I felt a bit flustered, but I got there in the end. The fourth question was about assumptions — I had to make my own assumptions to calculate different possible profits, something about bidding on land, with separate assumptions for winning the bid versus not winning it. The last question asked whether I'd actually bid on the land. I don't think there's a single correct answer here — you just need to be able to justify your own reasoning and talk it through with the interviewer.
Code was writing pseudocode, not in an actual testing environment. I think it was mainly testing data manipulation ability. I ended up writing everything with R dplyr-style logic — questions about imputing missing data, inserting, joining, that kind of thing. I deal with this kind of data regularly at my job, so overall it didn't feel too hard. Since I had extra time left over, I asked a lot of questions about what coding is actually required on the job. The interviewer said it's basically all Python now, which made me a little nervous since I'd been thinking in R the whole time — but he said that wouldn't affect my interview result, even though the job is clearly moving toward Python.
Role play was the same flight-delay case that's been around forever, already a known question on the forums. The interviewer gave me 15 minutes to look at slides, and I'd tell him to flip to the next page, and he'd flip it. Then I got about 20 minutes to present. I went with linear regression the whole way through. In the middle he asked what I thought of some of the charts — I said things like, using numbers for dates doesn't really make sense, since numbers don't capture the strong categorical relationship, and it should be changed to categorical data instead. At the end I had to give a recommendation. I'd prepared an opening beforehand — who I am, that I'd be walking him through this analysis today, roughly stating the motivation, then going through what I thought was important on each page, and finally giving a recommendation. It's basically the same routine I use for presentations at my current job — not sure if that helped, but the interviewer seemed to respond well to it.
Quant case was the classic dice-rolling problem from the forums — rolling three times. The first part was just roll three times straight through, the second part let you stop partway, and the third part let you stop partway but the second and third rolls each cost an extra dollar. I worked through it in order and had a lot of time left over, so the interviewer asked me some BQ questions and I asked him some too. With about ten minutes left, the interviewer told me to go rest up, and that was it — by then it was already late in the afternoon...
I didn't hear back right away after finishing. It wasn't until the following Friday afternoon that I got an offer call from HR. They said they only meet on Friday afternoons, so that's when results come out. Then the following Monday I got the formal offer. And with that, my job search wrapped up!!!! The whole process went really smoothly, felt pretty fast, and every interviewer was really nice.
Discussion
Loading comments…