Did the Yelp SWE loop last month while still employed, so I had to squeeze prep into evenings. Here's the honest rundown on the coding portion since I couldn't find a clean write-up anywhere.
For context I was interviewing at L4 (mid-level), backend track.
Online assessment: No OA in my pipeline actually. They went straight to a recruiter screen then a 45-min technical phone screen. Maybe the OA is for new grad / campus hiring only, or they scrapped it. Can't say for certain.
Technical phone screen: One problem, 45 minutes, HackerRank-style shared editor. The problem was a medium-difficulty array/string manipulation. Not the hardest thing I've solved but they want to see you narrate your thought process. I talked through two approaches before coding and the interviewer seemed to prefer that over someone who just starts typing.
Onsite coding rounds (2 of them): Both were medium-difficulty LeetCode problems. One was graph-adjacent (shortest path variant), one was more of a design-in-code problem where you had to implement a simplified version of something (like a mini cache). The cache one was more interesting and I spent more time on edge cases than raw algorithm speed.
Difficulty felt like: Yelp isn't trying to be Google. They're not throwing hard graph DP at you. But don't go in cold, mediums need to be clean and fast.
One gotcha: they cared a lot about test cases. After I wrote the code both times they asked me to walk through my test cases and explain why I picked them. This tripped up a friend who interviewed there around the same time.
Total prep I did: about 3 weeks, 1-2 hours a night. Focused on NeetCode 150, skipped the ultra-hard problems.
Got the offer, negotiated up a bit. Worth applying if you're targeting SF-area mid-level SWE roles in 2026.