Supply Chain Optimization with Bayesian Intelligence

Tony Sun, Ph.D., Lead Data Scientist, Lowe’s Companies, Inc

Abstract

Modern supply chains face persistent uncertainty in daily demand and lead times, especially at granular item–store–day levels. Traditional approaches, like exhaustive grid search, struggle to adapt efficiently in such complex environments. This session introduces a Bayesian optimization framework that intelligently learns from simulation noise to optimize safety stock decisions more efficiently. By continuously updating probabilistic beliefs about how safety stock levels impact key performance indicators—such as sales, in-stock rates, and GMROI—this method accelerates convergence on optimal strategies while enhancing transparency and robustness. Attendees will gain a practical understanding of how Bayesian techniques can improve supply chain performance, reduce computational overhead, and support smarter, data-driven inventory decisions in dynamic conditions.

About the Presenter

Tianxiao (Tony) Sun is a Lead Data Scientist and full-stack developer with deep expertise in AI and machine learning, particularly as applied to supply chain field. He earned dual bachelor’s degrees in mathematics (including Applied Mathematics) and Economics from Tsinghua University and holds a Ph.D. in Statistics and Operations Research from the University of North Carolina at Chapel Hill. With a strong interdisciplinary background, Tony has been an integral member of the Supply Chain Data Science team at Lowe’s, for six years. His experience spans the full spectrum of supply chain domain, including network flow forecasting, hierarchical clustering and classification, inventory replenishment simulation and optimization, and last-mile route and schedule optimization.