difficult and deep interview they asked about problems in scaling laws,current transformer architecture flaws,bias etc.there were total 3 rounds.You shouild have agood phd too for this role.Your research papers should be in RL,LLMs etc.

Was fairly quick -- spoke with the recruiter who expressed a lot of interest due to my very niche background. Did a substantial take-home and then had a meeting with a team member. Get the sense they aren't quite sure what they're looking for in this role.