Autonomous vehicles & robotics Data Scientist Interview Loop
Waymo is the company in this industry with enough recent reports on its data scientist interview. Zoox and Aurora are likely similar, but there aren't reports on them yet. The shape is an online test of SQL and statistics, a recruiter call, a live technical round on simulated fleet data, and an onsite. The onsite covers A/B test design, metrics, an autonomous driving case and behavioral. There's no LeetCode. The domain is what makes it hard. Safety events are rare, fleets are shared, and simulation has to be checked against the real world.
Practice by round
SQL over trips, vehicles and sensor logs, like trips per vehicle per day, disengagements per 1,000 miles, and sensor dropouts by vehicle and hour. Expect the duplicates and mismatched labels that real logs have.
Harder SQL at the right grain, like percentiles, fleet utilization and simulation pass rates. Then statistics coding: comparing Poisson rates, switchback experiments, and simulated versus real distributions.
| Question | Difficulty |
|---|---|
| Top 2 Sales Time Combinations | Medium |
| Sum Of Transportation Numbers | Medium |
| Sum Of Numbers | Medium |
| New Products | Medium |
| Car Part Price | Medium |
| Vehicle Gear Usage | Medium |
| Trip Duration Percentiles by Route Type | Medium |
| Simulation Pass Rate by Scenario | Medium |
| Fleet Utilization by Hour of Day | Medium |
| Miles Between Interventions by Month | Medium |
| Pickup Wait Time by Neighborhood | Medium |
| Mann–Whitney U With Ties | Medium |
| Two-Sample Kolmogorov–Smirnov Statistic | Medium |
| Sample Size for Two Proportions | Medium |
| Cancellation Rate by Weather | Hard |
Safety and rider cases, like safer software, ride comfort and pickup spots. Then mini-analyses on fleet data, including Software Release Safety, where you choose the safety metric yourself. Randomization comes first, because a shared fleet breaks a simple A/B split.
| Case study | Difficulty |
|---|---|
| Measuring Robotaxi Ride Comfort | Medium |
| New-City Robotaxi Launch | Medium |
| Wait Times After Fleet Growth | Medium |
| Is New Driving Software Safer? | Hard |
| Testing a Suggested Pickup Spot | Hard |
| Data project | Difficulty |
|---|---|
| Remote Assistance Rate | Easy |
| On-Time Pickup Test | Medium |
| Software Release Safety | Medium |
| Near-Miss Event Rates | Hard |
Rare events and their traps: zero collisions, overdispersed counts, Simpson's paradox across cities, and base rates for alerts.