Uber Machine Learning Engineer Interview Experience — 45 Min Coding Plus 15 Min Free-Form ML Design, Onsite Invite Next Day

The phone screen format was 45 min coding + 15 min design.

For coding I got the high-frequency "find robot" problem. I discussed the brute force solution with the interviewer, then how to precompute the distances in all four directions to speed it up, and finally wrote the optimized version. The interviewer didn't provide any test cases in HackerRank and didn't ask me to write any either. I only dry ran the sample cases that came with the problem out loud, and then we moved on to design.

The design part was very free form. Uber's own website guidelines say you will be asked about a challenge and the decision process from a project you worked on before, but the interviewer instead directly asked me where in the whole end-to-end UAIS flow, from the customer assigning a task to the driver completing it and uploading evidence, ML could be leveraged. ┓( ´∀` )┏ I gave a fairly shallow answer from three angles: using an LLM to collect leads from the customer, generating followable guidelines for the driver, and using ML to verify the quality of the evidence.

I got the onsite invitation the day after the phone screen.