Design a Drop-off Spot Selector
Company: Waymo
Role: Machine Learning Engineer
Category: ML System Design
Difficulty: hard
Interview Round: Onsite
Design an ML-driven decision system for an autonomous ride-hailing vehicle that must choose where to stop when a passenger is arriving at the destination and needs to exit the car.
The system should select a safe, legal, and convenient drop-off location near the requested destination rather than always stopping at the exact map pin. Describe:
- The product goal and what the system should optimize.
- The main constraints, such as safety, traffic rules, road geometry, passenger convenience, and impact on surrounding traffic.
- What candidate stopping locations you would generate.
- What data and labels you would use.
- What features you would build.
- What model or ranking approach you would use.
- How the online inference and decision pipeline would work.
- How you would handle uncertainty, edge cases, and fallback rules.
- What offline and online metrics you would use to evaluate the system.
- How you would iterate after launch.
Quick Answer: This question evaluates expertise in ML-driven system design for safety-critical autonomous vehicle decisions, covering real-time decision-making, geospatial reasoning, feature engineering, labeling and ranking models, data pipelines, and evaluation metrics, and belongs to the ML System Design domain.