Design a nearby-search system for autonomous vehicles. A query supplies a location and search criteria, and the system returns nearby vehicles that currently satisfy those criteria. ### Constraints & Assumptions - Vehicle locations change continuously and may arrive late or out of order. - Queries need a consistent definition of "nearby," such as a radius or top `k` by distance. - A temporarily stale location is possible and must be visible in the result metadata. - The exercise does not prescribe map geometry, fleet size, or a latency target; state the assumptions that drive your design. ### Clarifying Questions to Ask - Is distance straight-line, road-network travel distance, or predicted travel time? - Which vehicle attributes filter eligibility? - How stale may a location be before the vehicle is excluded? - Do queries need a snapshot, or is eventual consistency acceptable? - What are the update and query rates by region? ### What a Strong Answer Covers - Location ingestion, ordering, and last-known-state storage - Spatial partitioning and candidate retrieval - Precise filtering and distance refinement - Hot-region handling and repartitioning - Staleness, failure recovery, and observability - Privacy and authorization boundaries for vehicle data ### Follow-up Questions - How would predicted travel time replace geometric distance? - How would you handle a stadium-scale hotspot? - How would a vehicle removal race with a nearby query?