Research
My research focuses on helping firms make better dynamic decisions under uncertainty, real-world business constraints, and limited data availability, challenges that often require the development of new optimization and learning methodologies. I work at the intersection of stochastic optimization, approximate dynamic programming, and data-driven modeling. In particular, my research advances solution methods for large-scale Markov decision processes arising in emerging operations management applications by combining (i) modern approximation techniques, (ii) optimization and AI technologies, and (iii) high-dimensional sampling tools. The development of these methods is motivated by problems in inventory management, healthcare operations, and revenue management.
A broader goal of my research is to make real-world sequential decision models easier for non-expert business users to formulate, solve, and deploy. Rather than relying heavily on hand-crafted approximations, extensive parameter tuning, or user-driven reformulations, I study frameworks that automate key parts of the modeling and solution process while preserving strong theoretical guarantees. In particular, these self-adaptive frameworks incorporate mechanisms that inherently exploit problem structure, instance-specific data, and information revealed during the solution process, rather than requiring users to exploit and encode such structure in advance.
For a complete list of my published and working papers, please visit my research page.