Flatiron Health Data Scientist Interview Questions
& Process

This was a waste of time. It was a 1.5-hour HackerRank test with SQL and pandas questions. The SQL question was difficult, requiring Join, Case, and Coalesce. The pandas question involved calculating some KPIs. It wasn't technically difficult but easy to make mistakes on. Everything is optimized to auto-calculate a score and test many candidates, which just wastes candidates' time. The worst part is the outcome: no feedback from HackerRank on pass/fail, no score, nothing. Flatiron sends an auto-rejection email that isn't even formatted correctly, without mentioning any score. If you invest 1.5 hours, they should at least provide the automatically calculated score.

So they started me off with a take-home project that had some basic data science problems, basically asking potential clinical questions you could answer by looking at the data they provided. About two weeks later, we had a video call where they asked me stuff like SQL-query-like questions, which you know, could also be done with Pandas or other query languages, and then a couple of basic, high-level ML questions. I guess all these steps are pretty standard, but honestly, I'm not a huge fan of take-home projects. It really depends on how you approach it and how much you're willing to put in, you could easily spend days, not hours, on it, and the recognition you get won't match your effort unless you actually get the job. It just doesn't make sense for someone with years of experience, maybe out of a PhD program, to do coding projects like a fresh grad. It's kind of insulting. And what's even weirder is having someone with a Master's degree from a couple years ago interview a senior candidate. Do you really think their judgment reflects the candidate's abilities, especially when the candidate is way more experienced and knowledgeable? It's mind-boggling. It feels more like a dating service based on personal bias than a professional interview. I'm really disappointed with their process, the questionable standards, and the lack of actual feedback. Just a generic response saying it's not a match without explaining why. This whole interview culture needs to stop: biased, illogical, and a waste of time. Also, they should really talk to their audience at the right level. You don't send someone with a high school education to interview graduate students.

The interview process started with a phone screen with the hiring manager where we discussed what I'm looking for and my interest in the role. Then, there was a technical interview with the DS director about how I'd approach building a product from a given dataset. The final stage was a remote on-site with technical interviews, business scenarios, and behavioral questions. The whole thing took about 3 weeks. Some interviews felt a bit confrontational and overly enthusiastic about rapid prototyping. It felt like they were looking for an engineer more than a data scientist, though I know that's debatable.