x.ai Interview Questions
& Process

The recruiting team seemed very junior, struggling with complex sourcing and recruiting top-tier Research or Engineering talent. The interview process was shabby, lacking clarity on desired candidate profiles and facing difficulties closing candidates who also had offers from top AI companies. Questions weren't structured, and the overall recruiting team felt weak. I'd advise against joining xAI due to their derailed recruiting process.

I applied in September 2024. I passed GA and WA and got approved for the next steps super fast, within 24 hours. I even scheduled my interview for the very next day, so the whole process from GA to the interview took only 3 business days. It was the worst interview experience ever though. The interviewer asked a lot of questions but didn't give any feedback or even show any expression, which was really weird. There's no transparency about how they pick candidates. Honestly, they should just use AI to conduct the interviews.

Interviewing at XAI was tough but interesting. There were technical interviews covering AI basics, system design, and problem-solving. The interviewers knew their stuff and it was fun talking with them. They focused on real AI problems, which was cool.

Applied for the role and then got an email to sign up for CodeSignal and complete a General Assessment. Did the assessment the next day. The day after that, got an email stating, "The team has reviewed the results of your submission, and unfortunately, we will not be able to move forward with your application." This assessment was simpler than tasks I've actually done in this field for over 1.5 years, so I don't think I performed badly. I'll continue to believe this as long as there's no score on CodeSignal. It's odd to say they reviewed submission results when no assessment score is provided. I don't think I would be writing this if the score was visible, or if there was more general transparency. There might have been no flexibility in the answers (which isn't an efficient way to fix/train AI), or some other problem, but it's hard to tell since there are no pending results/no score for the assessment and no further explanation. Perhaps it's for the best.