This tutorial demonstrates CHORUS, a novel framework that uses Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) to generate Linear Programming (LP) code from natural language descriptions. Participants will learn how CHORUS generates accurate Gurobi solver code, connecting theory with practical use. The session will cover LP basics, the role of LLMs in code generation, and the unique approach of CHORUS, including its two-stage retrieval and cross-encoder reranking. Through interactive demos, attendees will understand the structure and applications of the framework. This tutorial aims to equip researchers and practitioners with skills to use CHORUS for optimization tasks and inspire advancements in automated code generation.

Target Audience: This tutorial is designed for PhD students, early-career researchers, academic professionals, and industry practitioners interested in automating Linear Programming (LP) code generation with LLMs. However, anyone with experience in basic Python programming and an interest in optimization or applied AI will also gain valuable insights. We do not expect prior expertise in LP or RAG and familiarity with LLMs is helpful but not required. Attendees should expect to come away not only understanding the CHORUS framework end-to-end but also having seen practical implementations that generate Gurobi solver code from natural-language problem descriptions.