Flipkart's engineering team details how they built the Central Planning Platform (CPP) to manage supply chain operations at massive scale — 500 million users, 150 million products, and 4 million daily shipments. The post explains two core lifecycle flows (inventory placement and customer delivery), the three structural challenges that drove the need for CPP (systemic interdependence, predictive precision vs. lead times, and dynamic scalability), and the two-layer architecture of the solution: a Forecasting Layer using ML/deep learning for multi-grain demand signals with hierarchical reconciliation, and an Optimization Layer using Operations Research to solve millions of constraints simultaneously. Key outcomes include a 90% reduction in planning cycle time (10 days to 1 day), 50% increase in 1-day deliveries, 50% reduction in unhealthy inventory, and 10% manpower reduction. The work was recognized as a finalist for the 2025 Franz Edelman Award in Operations Research.

10m read timeFrom blog.flipkart.tech
Post cover image
Table of contents
The Anatomy: How the Flipkart Supply Chain MovesLifecycle 1: The Inventory Journey (Getting It Close)Lifecycle 2: The Package Journey (Getting It to You)The Strategic Challenge: Planning at ScaleGet Piyush Vyas ’s stories in your inboxThe Solution: The Central Planning Platform (CPP)Introducing the 2 Technology Layers of the CPP1. The Forecasting Layer: The Eyes of the System2. The Optimization Layer: The Muscle of the SystemThe Impact: A Transformed FlipkartOperational VelocityEmpowering Our EcosystemConclusion & The Road AheadWhat’s Next?References

Sort: