Authored
CPG-Guided Context Improves LLM Vulnerability Patching for Taint-Flow Bugs
Research on improving vulnerability-fixing performance by giving language models better context from code property graphs.
Read paperResearch shelf
Papershelf
Research I have authored, studied, and found useful while working on AI systems, software security, developer tools, and distributed systems.
Research themes
AI agents and protocolsLLM security and reliabilityDeveloper toolsProgram analysisDistributed systems
Published work
Authored research
Authored
Agentic Content Protocol (ACP)
A whitepaper exploring a more sustainable protocol for interaction between AI agents and publishers.
Read whitepaperIdeas worth returning to
Reading list
Recommended reading
Why Language Models Hallucinate
A useful reference for understanding why language models produce plausible but unsupported answers.
Read paper