Hongbo Wen

I am a Ph.D. candidate in Computer Science at UC Santa Barbara, advised by Yu Feng and working closely with Dahlia Malkhi and Yanju Chen. Before UCSB, I received my B.Arch. from Tsinghua University.

My research brings programming languages and formal methods into LLM systems, aiming to build reliable and controllable intelligence. Viewing coding capabilities as a foundation for general-purpose AI, I develop formal verification, synthesis, analysis, and fuzzing techniques across both LLM lifecycles and critical systems:

Coding & Agentic LLMs: Specification-grounded analysis (SEMIA), fuzzing (SEFZ), runtime enforcement (VIGIL), and multi-agent vulnerability synthesis.
Compilers, Virtual Machines & Proof Systems: Synthesis-aided compilation (Tabby), VM fuzzing (BEAK), and constraint static analysis (ZKAP) over finite-field IRs.
Financial & Distributed Protocols: Attack synthesis over nonlinear arithmetic (FORAY) and verification-guided asynchronous protocol design (Thunderbolt).

My work has been supported by Google Security, a16z crypto, and Ethereum Foundation.

Email / Google Scholar / GitHub / CV

Publications

* Equal Contribution

ICSE'27 LLM Security Interpretable Security Analysis of Agent Skills Hongbo Wen, Ying Li, Hanzhi Liu, Chaofan Shou, Yanju Chen, Yuan Tian, Yu Feng pdf
NDSS'27 ZKP Program Fuzzing Revealing Semantic Gaps in Zero-Knowledge Virtual Machines via Obligation-Targeted Semantic Fuzzing Hongbo Wen, Jingyu Ke, Hanzhi Liu, Yanju Chen, Yu Feng pdf / bib / code
ACM CCS'26 Cryptographic Thunderbolt: Fast Asynchronous Off-Chain Bitcoin Transfers Hongbo Wen, Hanzhi Liu, Yanju Chen, Jingyu Ke, Dahlia Malkhi, Yu Feng pdf / bib / code
arXiv 2026 LLM Security VIGIL: Runtime Enforcement of Behavioral Specifications in AI Agent Skills Ying Li, Yanju Chen, Hongbo Wen, Bosi Zhang, Hanzhi Liu, Peiran Wang, Yu Feng, Yuan Tian pdf
arXiv 2026 LLM Security No Attack Required: Semantic Fuzzing for Specification Violations in Agent Skills Ying Li, Hongbo Wen, Yanju Chen, Hanzhi Liu, Yuan Tian, Yu Feng pdf
arXiv 2026 LLM Security Synthesizing Multi-Agent Harnesses for Vulnerability Discovery Hanzhi Liu, Chaofan Shou, Xiaonan Liu, Hongbo Wen, Yanju Chen, Ryan Jingyang Fang, Yu Feng pdf
arXiv 2026 LLM Security Your Agent Is Mine: Measuring Malicious Intermediary Attacks on the LLM Supply Chain Hanzhi Liu, Chaofan Shou, Hongbo Wen, Yanju Chen, Ryan Jingyang Fang, Yu Feng pdf
OOPSLA'25 ZKP Program Optimization Tabby: A Synthesis-Aided Compiler for High-Performance Zero-Knowledge Proof Circuits Junrui Liu, Jiaxin Song, Yanning Chen, Hanzhi Liu, Hongbo Wen, Luke Pearson, Yanju Chen, Yu Feng pdf / bib / code
ACM CCS'24 Financial Program Synthesis FORAY: Towards Effective Attack Synthesis against Deep Logical Vulnerabilities in DeFi Protocols Hongbo Wen, Hanzhi Liu, Jiaxin Song, Yanju Chen, Wenbo Guo, Yu Feng pdf / bib / code
USENIX Security'24 ZKP Program Analysis Practical Security Analysis of Zero-Knowledge Proof Circuits Hongbo Wen, Jon Stephens, Yanju Chen, Kostas Ferles, Shankara Pailoor, Kyle Charbonnet, Isil Dillig, Yu Feng pdf / bib / code

Zero-Day Discoveries

LLM Security	Filed 50+ zero-day security advisories on OpenClaw/ClawHub agent skills, covering credential leakage, RCE, PII exfiltration, and indirect prompt injection (e.g., clawhub#1911).
Financial	Discovered 10 zero-day logical vulnerabilities on BNB Chain DeFi protocols and synthesized end-to-end exploits reproducing $21M+ in historical attacks using FORAY (e.g., Discover).
ZKP	Discovered 14 zero-day vulnerabilities with BEAK (NDSS'27) while analyzing eight zkVMs: Ceno, Jolt, Nexus, OpenVM, Pico, Powdr, RISC Zero, and SP1, as well as additional zero-day vulnerabilities in widely-used Circom projects using ZKAP, including 0xPARC, Polygon-ZK, Axiom, iden3, and privacy-ethereum (e.g., circom-pairing).

Talks & Presentations

Feb 2026	Tabby: A Synthesis-Aided Compiler for High-Performance Zero-Knowledge Proof Circuits	SoCalPLS, USC, Los Angeles, CA
Feb 2025	Tabby: Automated Programming of Efficient Zero-Knowledge Proof Circuits	ETHDenver, Denver, CO
Feb 2025	Practical Security Analysis of Zero-Knowledge Proof Circuits	SoCalPLS, UCSD, San Diego, CA
Oct 2024	FORAY: Towards Effective Attack Synthesis against Deep Logical Vulnerabilities in DeFi Protocols	ACM CCS '24, Salt Lake City, UT
Aug 2024	Practical Security Analysis of Zero-Knowledge Proof Circuits	USENIX Security '24, Philadelphia, PA

Grants & Awards

2026	Finalist, Anthropic AI Safety Fellowship Top 5% of applicants
2024	Ethereum Foundation Academic Grant, Cybersecurity and Privacy Track Topic: Sentinel — Adaptive Counter-Attack Synthesis for Mitigating Onchain Exploits Yu Feng, Hanzhi Liu, Hongbo Wen
2023	Ethereum Foundation Academic Grant, Formal Verification Track Topic: Financial Model-Driven Attack Synthesis for DeFi Yu Feng, Yanju Chen, Hongbo Wen This grant led to the development of FORAY (ACM CCS'24).
2022	Academic Excellence Fellowship, UC Santa Barbara

Experience

2024 – Present	Co-founder & Principal Architect, Riema Labs Venture-backed ($12M raised). Led a 40-person engineering organization shipping production systems to 150K+ users through an AI-first engineering transformation; drove AI-era system design and quality control via explicit state-machine/schema contracts, automated contract lints, coverage ratchets, and AI review gates in CI.
2022	R&D Engineer, Veridise Inc. Audited production smart contracts and ZK circuits, uncovering 30+ critical vulnerabilities in widely used ZK systems and distilling recurring root-cause patterns that motivated ZKAP and FORAY.

Service

2025	USENIX Security '25 Artifact Evaluation Committee
2023 – 2025	Teaching Assistant, CS162 Programming Languages, UC Santa Barbara

Design inspired by Jon Barron