Building reliable systems - including the ones that think for themselves.
For the better part of a decade, I've helped fast-moving teams build, scale, and trust the systems their business runs on - and lately, that's meant building the infrastructure layer underneath AI agents and agentic tooling, not just the services around them.
I care as much about how something is built as whether it works. I value clarity over cleverness, judgment over hype, and work that holds up under pressure - whether that's a production rate-limiter, a Zero Trust network, or an MCP server connecting an agent to a real system.
- Start with the problem, not the tooling.
- Favor solutions that are simple to run and hard to break.
- Optimize for the long term - clarity, reliability, and trust.
- Treat AI tooling like any other production system: it needs to be secure, observable, and boring in the right ways.
Infrastructure and security for agentic AI systems - MCP integrations, self-hosted Zero Trust architectures, and abuse-prevention at scale. The unglamorous, load-bearing layer that makes AI products safe to run in production.
Open to conversations about AI infrastructure, agentic systems, and platform security.
