Background
Depthwise security expertise, now aimed at AI.
I have spent 10 years finding and fixing real vulnerabilities across high-stakes systems: security engineering, red-teaming, and full-stack penetration testing across highly-regulated attack surfaces.
My current work applies that operator mindset to frontier AI systems.
My theory of change is simple: turn sharp threat models into measurable research directions, find failure modes early enough to test and fix, then build the bridge from risk to action.
Security leadership
Senior offensive security work guiding generative and agentic AI security programs, adversarial testing, and remediation across complex enterprise systems.
Vulnerability research
Selected findings include a Bluetooth library supply-chain takeover vulnerability affecting 350M+ live devices and a mobile device lockscreen bypass affecting billions of devices.
AI security research
Heron AI Security Fellowship Research Lead working on distributed AI training threat models, interconnect telemetry, tensor recovery, and side-channel leakage.
Frontier red teaming
Built scalable LLM evaluation/red-team infrastructure and submitted findings on jailbreaks, reward hacking, misalignment, and dual-use risks.
Biosecurity bridge
BlueDot Impact Biosecurity Fundamentals, RosettaCon biosecurity session talk, and exploratory work on LLMs and biological design tools.
Strategy and governance
Completed the BlueDot Impact AGI Strategy course and separately explored governance questions around open-weight frontier models, multi-agent systems, and technical policy.