The screening for relevance team is 60 minutes. Half an hour answering questions regarding machine learning questions. This could include any machine learning techniques or terms. They would start asking you to talk about one related project and as you describe it further, they want you to dig deeper into its technical aspects. From then on, a lot of random machine learning jargon come up and you better have much deeper knowledge about them than just the concept. The second half an hour is a coding problem using the language of your choice. I found out preparing the LeetCode questions could be very helpful to answer such screening!

Phone screen then onsite. Phone screen: 1 coding question and 1 question regarding precision/recall. Onsite (6 rounds): 2 coding, 1 product design, 1 ML theory, 1 lunch, 1 behavioral. They asked a lot of questions in ML theory round. Recruiter's feedback: 1. I could not answer "what is '1' in F1 measure?" 2. I just describe superficially. Actually, I started thorough descriptions at the very beginning but the interviewer stopped me. Seemingly, the interviewers are biased toward people from Statistics or a certain country.