Design an API to let users sample from large language models efficiently, with good batching and request orchestration.

"I did a hard pre-assessment first, then the recruiter fast-tracked me because I already had other offers and got me through the rest of the loop in about a week. The onsite was five rounds for me: hiring manager, coding, company values, system design, and then another coding round. The most unusual part was the company values interview, which was run by nontechnical people and felt almost like a therapy session because they kept probing how I felt, not just what I did. The technical rounds were more practical than I expected, while the behavioral side was where I felt the real pressure. I finished the loop, but at the time of this conversation, I was still waiting on the outcome."