You are building the ML system behind a ride sharing application. The product must decide which nearby drivers to consider for a rider request and how to rank them so matching is fast, reliable, and efficient.
Design a scalable architecture for a ride-sharing application.
Candidate retrieval under real-time geo constraintsRanking and re-ranking for rider-driver matchingOnline feature serving and low-latency inferenceEvaluation under marketplace interferenceMonitoring drift, skew, and fallback behavior