Kevin Cheang

Researcher. UC Berkeley Ph.D.

headshot.jpg

SF Bay Area, USA

About

Hi, welcome to my homepage! I’m Kevin, a researcher and Applied Scientist at AWS AI Labs, where I train and evaluate LLM coding agents and develop SMT-based guardrails for safe agent tool use.

I received my Ph.D. in Computer Science from the University of California, Berkeley in August 2023, advised by Sanjit A. Seshia. During my Ph.D., I was a graduate student researcher in the Learn & Verify group and affiliated with the ADEPT/SLICE labs, working on formal verification, automated reasoning, and secure systems.

Before Berkeley, I received my B.Sc. in Combined Honors Computer Science and Mathematics from the University of British Columbia.

Research interests

  • Runtime monitoring and control of LLM agents, including SMT-based guardrails and outcome estimation.
  • Training and evaluating LLM coding agents, including building executable environments for reinforcement learning and long-horizon evaluation.
  • Formal methods and automated reasoning for safety-critical systems.

Selected publications

  1. arXiv
    Learning Context-Free Grammars for Grammar-Constrained Decoding via Declarative Agentic Programming with Guarantees
    Kevin Cheang ,  Geoff Hulette ,  Rahul Kumar , and 5 more authors
    Aug 2026
    PDF
  2. ASE
    Learning to Triage Vulnerability Reports from Program Analysis: An Empirical Study in Node.js
    Ronghao Ni ,  Aidan Z. H. Yang ,  Min-Chien Hsu , and 5 more authors
    In Proceedings of the 41st IEEE/ACM International Conference on Automated Software Engineering , Aug 2026
    PDF Website
  3. ASPLOS
    Lifting Micro-Update Models from RTL for Formal Security Analysis
    Adwait Godbole ,  Kevin Cheang ,  Yatin A. Manerkar , and 1 more author
    In Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2 , Aug 2024
    PDF
  4. CCS
    Cerberus: A Formal Approach to Secure and Efficient Enclave Memory Sharing
    Kevin Cheang* ,  Dayeol Lee* ,  Alexander Thomas , and 7 more authors
    In ACM Conference on Computer and Communications Security , Nov 2022
    * Equal contribution
    PDF
  5. CAV
    UCLID5: Multi-Modal Formal Modeling, Verification, and Synthesis
    Elizabeth Polgreen ,  Kevin Cheang ,  Pranav Gaddamadugu , and 5 more authors
    In The 34th International Conference on Computer Aided Verification (CAV), August 2022 , Aug 2022
    PDF
  6. CAV
    The Move Prover
    Jingyi Emma Zhong ,  Kevin Cheang ,  Shaz Qadeer , and 6 more authors
    In Proceedings of the 32nd International Conference on Computer Aided Verification (CAV ’20) , Jul 2020
    PDF
  7. CSF
    A Formal Approach to Secure Speculation
    Kevin Cheang ,  Cameron Rasmussen ,  Sanjit A. Seshia , and 1 more author
    In 32nd IEEE Computer Security Foundations Symposium (CSF ’19), Hoboken, NJ, USA, , Jul 2019
    PDF
View all publications

Awards

Services

Program committee / paper reviewer

  • 2026: ESORICS
  • 2025: ESORICS
  • 2024: ECOOP, TACAS, VMCAI
  • 2023: POPL - PriSC
  • 2021: CAV, FMCAD
  • 2019: TACAS

Artifact evaluation committee

  • 2024: ICCPS, USENIX, ECOOP, TACAS
  • 2023: CCS, ICCPS, CAV
  • 2022: CAV

Teaching

University of California, Berkeley

  • CS 164, Programming Languages and Compilers (Fall 2020)
  • EECS 219C, Formal Methods: Specification, Verification, and Synthesis (Spring 2019)

University of British Columbia

  • CS 221, Basic Algorithms and Data Structures (2014)
  • CS 213, Introduction to Computer Systems (2014)
  • CS 121, Models of Computation (2013)

Contact

To contact me, please find me on LinkedIn!