BenchSci Interview Questions
& Process

I applied for a UK-based Senior ML Engineer role. Started with HR screening, then a technical call. The technical interview felt off, with the interviewer mentioning 'vibe coding' and trying to make it very conversational, agreeing with me a lot. He assured me I'd pass and discussed the next steps. Then I got a rejection email saying they wanted someone more senior, which didn't align with the earlier chats. It felt like a performance rather than a real evaluation, lacked transparency, and showed poor company culture.

You start with a zoom screening with the Recruiter, and if that goes well, the hiring manager will talk to you. If the hiring manager thinks you might be a good fit, you'll get case studies to work on. These case studies usually have a week to prepare and you present them to a group of 3-4 people from the current team or from other teams.

The HR round was great. The first technical interview seemed like the interviewer wasn't really engaged and lacked experience. They just had a basic list of questions. It was a 45-minute video call to discuss my work history and experience, behavioral and situational questions, and my personal values. This felt impossible to cover in the time given, especially since 15 minutes were spent explaining the company and teams, leaving only 30 minutes for the actual interview. Some design questions were confusing as they didn't have clear answers and it wasn't a conversation at all. It was very formal, with me answering and the hiring manager just saying "thank you for that" after each response. They didn't seem interested in my examples, even saying it was okay not to use them.

I applied online. A recruiter reached out to me and scheduled the first two interviews: Python Refactoring and System Design. I got an email from the interviewer a few minutes before the interview with a zip file which needed refactoring (they assume you have an environment set up on your machine without communicating this to you in advance). During the interview, he asked to open the project in an IDE and start refactoring. It is a very unstructured and open ended interview. I came out of it thinking I did very well given the unclear instructions and the fact that you can't finish it all within 45 minutes. The system design interview was also conducted in an unstructured way. The interviewer came to the meeting unprepared and took them a while to set up a whiteboard where he could explain the problem. The problem is very specific to BenchSci domain. I got a rejection email a few days after saying that they could provide more feedback on my performance if I ask for it and they will provide it to me within max 2-3 days. I asked for feedback and did not hear back for a week despite following up. After a couple of weeks, the recruiter responded with some poorly written feedback that did not make sense. They only shared the 'areas for improvement' with a few vague bullet points which out of 5, I actually had done 4 of them during the interview (unjust and lousy feedback!). This tells me the interviewers can get away with anything as the process is not regulated. They can let in their buddies and block people they have biases against. I'm glad this didn't work out for me as I wouldn't have wanted to work with these people.