I emailed the recruiter directly and he called me back very soon after. we had a chat about my career so far, my main motivation to apply and so on. Next was a video call with the head of data science. this was not very technical but also mainly about my motivation, skills, experience,etc. as well as him explaining the exact role and projects that the job includes. next was a technical interview with a backend engineer and another data scientist. it was one these typical software engineer live coding interviews, where they ask you to code something and if you are lucky enough to have memorized the answer, you would be able to answer. so to me it seemed as if they were looking for someone who has explicitly practiced for coding interviews, not necessarily for a good data scientist. even when i coded a question correctly, they asked me to do it again but without using a method provided by a library. which makes no sense to me, because every uses libraries and the library's implementation of that method is probably better than my own anyway, so why make someone implement something on their own, when in the job this is rarely needed. instead of letting candidates google style live coding tests with a super narrow scope under circumstances that would never occur in real life, just let them finish a task or assignment over the duration of some days and let them present their results and evaluate their qualifications based on that.