Presentation: From Reactive to Predictive: How ML-Based Carrier Risk Scoring Reduced Pickup Defects by 35%
Speaker: Debanshu Sharma, Senior Supply Chain Manager, Amazon
The session will cover a machine learning framework built and validated across 150,000+ loads and 1,600+ carriers that generates pickup risk scores at the carrier level, shifting the operational model from reactive exception management to proactive intervention. Results include a 35% reduction in pickup defects in the high-risk carrier segment, 85% model accuracy, and $40M+ in projected annual savings.
Speaker
-
Debanshu SharmaSr. Supply Chain ManagerAmazonDebanshu Sharma is a supply chain and transportation strategy expert with 15+ years of experience spanning freight network design, carrier performance optimization, predictive analytics, and large-scale logistics operations. He currently leads transportation strategy initiatives at Amazon Transportation Services, where his work spans carrier performance optimization, machine learning-based logistics analytics at scale, and network design optimization across one of the world’s largest transportation networks.
Debanshu developed the Predictive Reliability Index (PRI), a machine learning framework validated on 150,000+ shipment records across 1,600+ carriers that generates pre-departure carrier risk scores — yielding a 35% reduction in pickup defects and 85% model accuracy. The framework is the subject of a peer-reviewed academic paper co-authored with Dr. Amulya Gurtu, Department Head at Texas A&M University, accepted for presentation at the 2026 INFORMS Annual Meeting in San Francisco under the Transportation Science and Logistics track.
His published work spans Supply Chain Management Review, CSCMP Supply Chain Xchange, Inbound Logistics, SupplyChainBrain, and Supply & Demand Chain Executive. He serves as a peer reviewer for IEEE Transactions on Engineering Management and Taylor & Francis journals.
Debanshu holds an MBA from the University of Michigan Ross School of Business and an MS in Mechanical Engineering from SUNY Buffalo. He is based in Metro Detroit, Michigan.
