I went through five interviews, alternating between technical and HR/soft skills after each round. The process took about a month.

The machine learning questions were quite basic and revolved around live-coding exercises where you implemented parts of algorithms, like k-NN, live.

The final interview was a bit unusual. I had been a finalist with one other person, and the team leader conducted the interview. I answered questions that didn't have concrete answers, meaning if you were on theme, you could argue many possible solutions for a specific problem.

The interviewer did not like my approaches and advocated for the use of a certain paper they had co-authored. It was a rather disappointing end to an otherwise good, albeit tiring, interview process. Ultimately, I wouldn't want to work for someone who imposed their views on me, especially on problems that are open research questions.

Several questions arose across different interviews.

I was asked to implement kNN.

For the final interview, I was asked to select an algorithm to deal with video scene identification.

The HR also asked standard questions common in the interview process of a large company, such as describing a time when I explained a technical matter to a non-technical audience.