I was interviewing for a DS role while still employed elsewhere. The overall process was decent, and after the interviews I felt pretty good about how it went, but in the end I still got a fail. Two rounds said my answers weren't detailed enough. Personally I think I was already quite detailed, and there wasn't a single question I couldn't answer or got wrong. Has anyone gotten similar feedback before and can tell me how to improve?
The first round was with a hiring manager talking about whether my background was a fit. It mainly asked about projects from my current job, and it went pretty deep, touching on some ML knowledge.
Then there was a take-home that was supposed to be under 6 hours, but I spent at least 4 times 6 hours on it. It was about predicting the target users of a product. The prompt was very open-ended, there was a lot you could do with it, and it really depended on how deep you wanted to go.
The HM round and the Business Partner round both wanted you to bring in real examples from your work.
There was a presentation round, where you present the results of your take-home. You need to put together new slides for it, and you also need to really understand what you actually did, because the Q&A afterward has modeling-related questions.
The stats round mainly covered two parts: one was digging deep into projects I'd done, and the other was them giving a topic and having me work through it live. It felt more like a modeling-heavy case study.
The case study round was the usual format, related to evaluating a new product, or estimating the trend of some metrics.
The SQL question itself was simple, but it was embedded inside a case, so really it should also count as a case study round. Mainly it was about designing a metric for the problem, then writing SQL. 80% of the time went into designing the metric and analyzing why the metric looked abnormal.
Discussion
Loading comments…