In recent years, AI-based Software Engineering has experienced remarkable growth, with significant advancements in code generation, test generation, bug detection, and bug repair. However, current Code Language Models (Code LLMs) often lack a deep understanding of code semantics, resulting in performance issues and a general lack of trust in their outputs.

In this talk, I will explore how we can design, implement, and utilize AI models that effectively leverage the semantics of source code. By understanding and applying code semantics, we can create models that outperform state-of-the-art LLMs, often with smaller, more cost-effective models. Join me as we uncover how semantic-aware AI models can revolutionize software engineering, providing robust, dependable, and advanced tools for developers.

Baishakhi Ray is an Associate Professor of Computer Science at Columbia University. She works at the intersection of AI, Software Engineering, and Security. She received her Ph.D. degree from the University of Texas, Austin. She has received the IEEE TCSE Rising Star, NSF CAREER, IBM faculty, and VMWare Faculty awards. She is also an Amazon Visiting Academic. Her research has been acknowledged by many Distinguished Papers awards including FSE’17, ASE’22, ISSTA’23, the ICSME Most Influential Paper award, and has also been published in CACM Research Highlights and has been widely covered in trade media.