I am a research scientist at a FAANG. The interview process we went through is much less demanding in the CS part, as the domain knowledge is the focus (yet we

I am a research scientist at a FAANG. The interview process we went through is much less demanding in the CS part, as the domain knowledge is the focus (yet we still struggle to find good candidates that pass). But I've noticed that those who passed the coding interviews (easy to medium leetcode style) write much more efficient, maintainable, and concise code for the algorithms, experiments and analysis they implement. This results in much higher productivity, even if they have to do more catch up on domain-specific knowledge. The blog post mentions he works on "machine learning" projects, which in my experience benefits significantly from having leetcode-style algorithmic knowledge (assuming you are not just calling blackbox.infer() all day).