Problem
Design a nearby POI service that returns ranked results near a user's location from 200M records
- Narrow candidates efficiently across 200M POIs without brute-force distance scans
- Guarantee accuracy at cell boundaries by expanding searches to neighboring spatial zones
- Scale hot metro traffic through prefix-based sharding and cache strategies for skewed demand
Key Test Points
- Geohash indexing
- Neighbor-cell expansion
- Prefix-based sharding
Spot an issue or have a suggestion?