About Me

My Approach

I believe that the best AI systems aren't just powerful—they're transparent, observable, and safe by design. Too often, AI development treats safety and ethics as afterthoughts. I work to embed these properties into the architecture itself.

My work spans from hands-on system building to theoretical research. I design multi-agent architectures that can coordinate safely, evaluation frameworks that catch failures before deployment, and observability patterns that make AI reasoning legible to humans.

What I Work On

Production AI Systems

Building scalable, reliable AI systems that work in the real world. This includes architectural design, infrastructure decisions, deployment strategies, and the practical engineering required to take AI from prototype to production.

Multi-Agent Architectures

Designing coordination frameworks for systems where multiple AI agents need to work together. This requires careful attention to communication protocols, authority hierarchies, and failure modes that emerge from agent interactions.

AI Safety & Evaluation

Deep evaluation of AI systems for safety, reliability, and alignment. This includes red-teaming, benchmark development, systematic failure analysis, and research into how AI systems can be made more robust.

Observable AI

Developing patterns and practices for making AI reasoning visible and interpretable. Not just logging outputs, but tracing the full trajectory of decisions, considerations, and alternatives that lead to each action.