Design an algorithm and high-level system for recommending a pickup point in a ride-hailing app. A passenger requests a ride from an approximate city location, such as inside a building or near a large intersection. A driver is at a known GPS location. The system should recommend a safe and practical pickup point for the passenger to walk to and the driver to reach. ### Constraints & Assumptions - You have access to map data, road graphs, pedestrian paths, routing/ETA services, and real-time traffic when available. - The passenger GPS pin may be noisy or inside a building. - The driver must stop legally and safely. - The pickup point should balance passenger walking time, driver ETA, safety, reliability, and user experience. - Present the selected point clearly to both passenger and driver. ### Clarifying Questions to Ask - What city environment is in scope: downtown, suburb, airport, campus, or mixed? - Is passenger walking ability or accessibility known? - What is the maximum acceptable walking distance? - Do we know legal stopping zones, curb access, entrances, and historical pickup success? - Is the objective minimizing total time, cancellations, unsafe stops, or support contacts? ### What a Strong Answer Covers - Snapping passenger and driver positions to map/road/pedestrian networks. - Candidate pickup-point generation near the passenger. - Filtering unsafe or illegal candidates. - Scoring candidates using driver ETA, passenger walking ETA, safety, confidence, and historical success. - System architecture with client apps, backend, map service, routing service, pickup-point service, logging, and model improvement. - Presentation in both apps and fallbacks for low-confidence cases. ### Follow-up Questions - How would you handle airports or malls? - How would you learn better pickup points from historical data? - How would you support passengers who cannot walk far? - What metrics would you use to A/B test the pickup algorithm?