Failed an Uber L5 SWE loop two months ago. Writing this out partly for myself and partly because I couldn't find anything this specific when I was prepping.
The loop was: 1 phone screen (LC medium, 45 min), then 4-round virtual onsite: 2 coding rounds, 1 system design, 1 behavioral.
Where I think I failed:
Coding. Got a graph problem in one round that I recognized but fumbled the edge cases. I jumped to the solution too fast without talking through the problem structure. The interviewer tried to redirect me twice and I didn't pick up on it. In hindsight: Uber interviewers seem to use these redirects as signals of coachability. I plowed ahead. That probably hurt.
System design. I over-engineered. Asked to design a rate limiter, I went full distributed Kafka-backed monstrosity in the first 10 minutes. The interviewer kept asking 'what would you simplify here' and I kept adding complexity. At L5 Uber specifically, they want you to identify a simple correct design and then layer in scale considerations. Not start at maximum scale.
Behavioral. This one I thought went well, but the feedback I got (they gave some, which I appreciated) mentioned that my answers were 'high level.' I think I was telling stories at the project level rather than my individual contribution level. They want to understand what YOU specifically did, not what the team did.
Feedback I received: passed coding (I guess the graph round wasn't as bad as I thought), borderline on system design, not passed on behavioral. Surprising.
What I'd do differently: slow down on coding, stay leaner on system design for the first 15 minutes before scaling, and practice behavioral answers with 'I did' not 'we did' framing.
Might reapply in 6 months. Uber's re-application window is usually 6-12 months depending on role.