Two phone screens, coding with AI. My feeling is that you basically just need to show your ability to interact with AI — let the AI write all the code, and your job is mainly to review it, ask it what algorithm to use, how to design the functions and classes, and then once the code is done you comment on the problems with what it wrote. That's basically fine. I don't remember the actual problem.
System design: a pretty standard interview — design a metrics monitoring system. He simplified the metrics down to just counts, so basically the two chapters in Alex Xu's book on ad click counting and metrics monitoring were enough to answer it.
Onsite was four rounds. Two of them were behavioral, one was a project deep dive. The remaining round was an integration round on a bike map — I only finished one question, because I was writing it in Java, which was more verbose, but the interviewer said finishing one question was fine. He said if it had been Python, the bar would have been higher.
In the end I didn't pass. The feedback was that there was no problem on the technical side, but I didn't answer the behavioral questions well — the content leaned too much toward execution and didn't show staff-level scope. Anyway, I just have to keep improving.
Discussion
Loading comments…