From Code Generation Towards Software Engineering: Advancing Code Intelligence with Language Models

This Allen School Colloquia Series lecture explores the evolution from basic code generation to comprehensive software engineering using large language models. Discover how PhD candidate Yangruibo (Robin) Ding from Columbia University advances code intelligence with LLMs by enhancing their symbolic reasoning for program semantics and global reasoning for software dependencies. Learn about his research at the intersection of Software Engineering and Machine Learning, where he trains LLMs to generate, analyze, and refine software programs while building evaluation benchmarks. The talk outlines a vision for approaching full-stack automation in software engineering in a trustworthy manner. Ding's award-winning interdisciplinary research spans top conferences in software engineering, programming languages, natural language processing, and machine learning, including an ACM SIGSOFT Distinguished Paper Award and an IEEE TSE Best Paper Runner-up.