Sorin Lerner
Sorin Lerner is a computer scientist whose research applies AI to program proofs and programming tools. In 2003, he coauthored a method for automatically proving compiler optimizations sound; in 2020, he coauthored Proverbot9001, which used machine learning to generate software correctness proofs. He studied computer engineering at McGill and earned a computer science PhD at the University of Washington before joining UC San Diego in 2006. In 2024, he coauthored CoqPyt, a Python tool for working with Coq proofs, and a study of live programming for validating AI-generated code. He is dean of Cornell Bowers and a professor of computer science, with current work on LLM-based program proofs and human-centered programming environments.
- Now at
- Cornell University
- Works
- 192
- Citations
- 4,246
- h-index
- 32
- Badges
Publication numbers from OpenAlex, as of 27 October 2026.
Timeline
-
2003
Correct compiler optimizations
With Todd Millstein and Craig Chambers, he develops Cobalt, a domain-specific language for compiler optimizations, and a method to prove those transformations sound. Their PLDI paper wins the 2003 Best Paper Award.
-
2006
Joining UC San Diego
He grows up in Montreal, studies computer engineering at McGill and earns a computer science PhD at the University of Washington. In January 2006, he joins the Jacobs School's Computer Science and Engineering faculty.
-
2018–2019
A teaching award
At UC San Diego, he receives the Jacobs School of Engineering's CSE Teacher of the Year award for 2018–2019.
-
2020
Proverbot9001 generates proofs
In 2020, he coauthors Proverbot9001, a proof-search system that uses machine learning to produce software correctness proofs in interactive theorem provers.
-
2024
Coq proof navigation
As a coauthor of the 2024 CoqPyt paper, he presents a Python tool for interacting with Coq and extracting rich premise data. The tool supports work on LLM-based proof synthesis and repair.
-
2024
Checking AI-generated code
At CHI 2024, he coauthors a study that builds a Python editor combining an AI programming assistant with live programming. In a 17-person experiment, the team finds lower cognitive load for some AI-code validation tasks.
-
2025
Rango for proof synthesis
He coauthors Rango, an automated tool for Coq proof synthesis, in 2025. It retrieves relevant premises and similar proofs at each step, and receives an ACM SIGSOFT Distinguished Paper Award at ICSE 2025.
-
Since November 2025
Dean of Cornell Bowers
He is dean of the Cornell Ann S. Bowers College of Computing and Information Science and a professor of computer science, appointed in November 2025. Before joining Cornell, he serves as chair of UC San Diego's Department of Computer Science and Engineering and helps guide the launch of its artificial-intelligence major and online master's program in data science. His recent work uses AI, machine learning and LLMs to build program proofs, and applies human-centered design and user studies to improve programming environments.
Career
- Since 2025Cornell UniversityDean, Cornell Ann S. Bowers College of Computing and Information Science; Professor of Computer Science
- From 2006University of California, San DiegoComputer Science and Engineering faculty member
- University of California, San DiegoChair
Education
- McGill UniversityComputer Engineering studies
- University of WashingtonPhD in Computer Science
Sources and credits
Photo: We have no published photo of Sorin Lerner yet. Send us a free-licensed photo.
- Automatically Proving the Correctness of Compiler Optimizations
- Faculty Profiles | Jacobs School of Engineering
- Congrats Sorin Lerner: CSE’s 2018-19 Teacher of the Year
- Generating Correctness Proofs with Neural Networks
- CoqPyt: Proof Navigation in Python in the Era of LLMs
- Sorin Lerner
- Validating AI-Generated Code with Live Programming
- Rango: Adaptive Retrieval-Augmented Proving for Automated Software Verification
- ACM SIGSOFT Distinguished Paper Award at ICSE 2025
- Sorin Lerner | Cornell Bowers




