Waymo Software Engineer Interview Guide

What stands out most across Waymo's software engineering interviews is how consistently candidates are caught off guard by the breadth of the loop. The pattern we see consistently is candidates preparing for a standard DSA-heavy SWE process and instead encountering ML system design, robotics domain questions, and autonomous driving context — sometimes in the same round as a graph traversal problem. One candidate described being asked about Vision Transformers and designing an ML system for pedestrian crosswalk prediction; another got ML questions in what was labeled a standard SWE loop. If you're interviewing here as a software engineer, the autonomous driving domain isn't background color — it's part of the actual evaluation.

The coding questions themselves show a clear pattern: graph problems, particularly BFS/DFS on 2D matrices, appear repeatedly, but Waymo's versions tend to have a twist that makes template solutions insufficient. Candidates who got the core idea but didn't push toward a complete, edge-case-hardened solution consistently reported that it wasn't enough. One candidate noted that even with interviewer hints on a recursive matrix problem, the bar was clearly on correctness and completeness, not just approach. We've also seen reports of a data-wrangling style coding task that felt more like navigating an underspecified codebase than solving a LeetCode problem — so the format can shift significantly depending on the team.

The process also has a real inconsistency problem that candidates should mentally prepare for. One interviewer was described as not knowing what they were supposed to ask and throwing out random technical questions for 30 minutes. Another changed an hour before the interview. The variability isn't a sign of a broken process so much as a signal that different teams at Waymo are running somewhat different loops — which means your experience will depend heavily on which team you're interviewing with and who shows up in the room.