Writing
Continuous Assurance Fabric Reference Architecture (CAF-RA)
A reference architecture for governed, scalable continuous assurance at mission tempo, with policy-governed swarm adjudication, replayable evidence, and enforced remediation closure.
Don't Put Frontier AI Under ITAR: A Blunt Instrument for a Dual-Use Reality
A policy memo arguing against sweeping ITAR treatment for frontier AI and proposing a targeted control framework centered on dual-use export controls, defense-unique controls, and enforceable operational provenance.
Stand Up Delegated Autonomy Directorate (DAD) as a Joint Control Plane Authority
Decision memo recommending immediate standup of a Delegated Autonomy Directorate (DAD) as a joint authority to accelerate trusted delegated autonomy at mission tempo.
Agent Control Plane Reference Architecture (ACP-RA)
A reference architecture for governed, scalable agentic autonomy—single agents and swarms—aligned to DoD CIO patterns (Zero Trust, ICAM, CNAP, DevSecOps, cATO) and designed for contested/degraded operations.
Notes: Research that shaped ACP-RA (agent security, tool use, evaluation)
Reading notes on prompt injection, tool-use at scale, and execution-based evaluation that drove ACP-RA design choices (gateways, envelopes, evidence, anti-replay, and upgrade discipline).
From AI Force Multiplication to Force Creation
A white paper on agentic autonomy, trust scopes, and strategic imperatives for defense.
From PDFs to Pull Requests
Code-as-Policy: transforming Department of Defense policy workflows with DevSecOps, version control, and continuous verification.