Baishakhi Ray

Baishakhi Ray is an Associate Professor at Columbia University leading research at the intersection of AI and Software Engineering. She serves as an Amazon Scholar collaborating with Amazon's Agentic AI team and heads the ARiSE Lab focused on making code smarter, safer, and more collaborative through neurosymbolic AI techniques.

Her research centers on neurosymbolic approaches that blend machine learning with formal reasoning to enhance software reliability, security, and developer productivity. Key areas include AI for Code, language model development for code generation/transformations, and benchmarking frameworks for evaluating LLMs in software engineering contexts. She pioneers techniques for building trustworthy code through semantics-aware language models and agent-based systems.

Recent publications (2024-2025) reveal a strong emphasis on neurosymbolic AI applications across the software lifecycle - from code generation (EditLord, Cycle) and vulnerability detection (Primevul) to test repair (UTFix) and kernel debugging (Kgym). Her work consistently bridges neural and symbolic paradigms to address fundamental challenges in code reliability and developer tooling.

Major recognitions include:

• IEEE CS TCSE Rising Star Award
• NSF CAREER Award (2019) and IBM Faculty Award (2019)
• Multiple ACM SIGSOFT Distinguished Paper Awards (2023)
• NAACL 2025 Oral Presentation and ICSME 2023 Most Influential Paper
• VMware Early Career Faculty Award (2020)

She mentors a dynamic research group including PhD students Yangruibo Ding (recent UCLA faculty hire), Vikram Nitin, Ira Ceka, Jinjun Peng, and Alex Mathai. Research is funded by:

• Government: National Science Foundation
• Corporate: Google, IBM, Amazon, Capital One, RedHat
• Institutional: Columbia Provost’s Grants for Junior Faculty

The ARiSE Lab maintains active industry collaborations through Amazon Scholar role and corporate partnerships, driving neurosymbolic AI innovations from theoretical foundations to real-world software engineering applications with ongoing projects in coding agents and secure code transformation.