The interview process is quite lengthy. This is primarily because the next available time slot for interviews is often more than two weeks away. I didn’t pass the homework assignment review stage. In that interview, candidates are typically asked about general machine learning knowledge, with the expectation of specific answers.

There are in total 4 interview rounds. The first is a typical Python coding test where you have to solve three (easy - medium) Leetcode questions. The second is a Machine Learning specific interview, where you are asked basic questions regarding (modern) Machine Learning, i.e. Gradient Descent, Neural Networks, Optimizer etc. For that the interviewer discusses a take-home assignment that you have to submit prior to the interviews. The third is a 30-minute presentation of a recent AI paper of your choice. Afterwards, there is a discussion with the audience where you have to answer questions. The questions might also be somewhat unrelated to the presentation itself. The fourth is directly after the presentation, which is a short "culture screen". After the interviewer explains a little bit in more detail what the company does, he asks for your desired wage, your desired location (Japan/Abroad), whether you need sponsoring etc. In the end he also asks a question regarding your personality.