Democratizing Optimization for All: A Co-Pilot for Decision Making

Hack Session

About the session

Optimization is one of the most powerful tools for decision making, yet it remains inaccessible to most domain experts because turning real-world goals and constraints into formal mathematical models requires specialized expertise. This talk explores how large language models can change that. We present an AI co-pilot that translates natural-language problem descriptions into optimization models, code, and solution workflows, while keeping humans in the loop for oversight and refinement.

For dynamic, process-oriented problems, the framework further uses LLM-generated digital twins and agentic reinforcement learning to discover effective decision policies. By lowering the barrier between domain knowledge and advanced optimization, this approach aims to make optimization usable by everyone, not just experts, and to reshape how complex decisions are modeled, solved, and deployed.

Speaker

Download Brochure