At peak demand, Rapido processes nearly 6.5 million ride requests daily across bikes, autos, and cabs. In an exclusive conversation with AIM, CTO Srivatsa Katta described the platform as a highly distributed system designed to match riders and drivers instantly, even during traffic spikes or infrastructure failures. Backend Built for Scale Rapido's backend runs on hundreds of microservices hosted in the cloud, with each service responsible for specific tasks such as location search, ride dispatching, pricing signals, or ETA calculations. The core principle guiding these systems is resilience. Every component runs in clusters and is backed by automatic scaling and redundancy so that failures in one part of the system don't affect the rest. When a user requests a ride, the system immediately performs multiple geographical queries to identify nearby drivers, evaluate potential routes, and determine whether the trip is feasible. Ride requests are then sent to available captains, along with details such as the pickup location, drop destination, and estimated earnings. Drivers can review these details before accepting a ride, a design choice Rapido says helps maintain transparency within the marketplace. However, matching riders and drivers quickly is only part of the challenge. The company continues to optimise two critical metrics: how quickly a driver accepts a ride and how fast they reach the passenger. To improve these decisions, the platform relies heavily on data. Since its launch in 2015, Rapido has recorded more than three billion rides. This dataset feeds machine learning systems that help predict routes, estimate arrival times, and improve ride dispatch decisions. Historical data is combined with real-time traffic information from mapping providers. Together, these signals help the system estimate more accurate ETAs and determine which drivers should receive ride requests. Mapping itself presents a unique challenge in India. Road closures, ongoing construction, and unpredictable traffic behaviour often make standard navigation rules unreliable. In many cases, how people actually drive differs significantly from how roads are mapped digitally. To address this, Rapido does not depend on a single mapping provider. Instead, it combines multiple mapping services, open-source data sources such as OpenStreetMap, and its own historical ride data. Even with sophisticated algorithms, however, the final decision always rests with the driver. Captains often evaluate whether a ride is economically worthwhile before accepting it. Factors such as pickup distance, drop location, and the likelihood of securing another ride after the trip all influence their decisions. |
Комментариев нет:
Отправить комментарий
Примечание. Отправлять комментарии могут только участники этого блога.