Yelp ML Engineer Interview Guide

Candidates who interviewed for Yelp's machine learning roles describe a process that leaned more on coding fundamentals than on ML-specific material. The ML Engineer candidate said they went in expecting mostly ML and found much of the process was algorithms.

Both reports open with a coding stage. The ML Engineer candidate describes a HackerRank assessment with a SQL query and a LeetCode-style problem solvable with a heap. The Senior ML Engineer candidate describes an offline coding challenge followed by a recruiter chat in which the recruiter laid out the next steps. Both then describe a live coding interview. The ML Engineer candidate did it on HackerRank and suggests talking through your approach before typing and being ready to explain time complexity. The senior candidate describes theirs as interactive, with a Yelp engineer.

The final stage was the longest part. The senior candidate describes a roughly 4-hour back-to-back panel with two behavioral interviews, one system design interview, and one live coding interview. The ML Engineer candidate describes four separate final interviews covering coding, system design, and behavioral questions, and says system design is where they would prepare most. The senior candidate found the system design interviewer dismissive and overly critical, though a manager in one behavioral conversation was kind, so it helps to practice defending design choices under pushback.

The senior candidate's live coding question asked for Jaccard similarity between sentences, which means tokenizing and handling set logic cleanly. Practice similar string and set problems, heap-based problems, and SQL under time limits. Both candidates reported no offer. The senior candidate said the technical difficulty was manageable but the length of the panel made it hard to stay sharp, so a few long mock sessions are worth the time.