About

I spent 21 years as a sworn law enforcement officer, retiring as a sergeant. That means two decades of using the tools this industry sells: the cameras, the plate readers, the records systems, the video that did or didn't hold up when it mattered.

Much of that career was spent on major investigations and covert surveillance: building covert hides, running video surveillance, and standing up quick, hasty deployments of covert and overt camera platforms to capture intelligence or evidence on long-running, sensitive, high-value cases. When a platform failed on one of those, you didn't lose a data point. You lost weeks of work, and sometimes the case.

Most of that technology was built by people who never had to depend on it at 2 a.m. You can tell.

Today I work on the other side of the counter. My day job is leading law enforcement business development for an AI video analytics company. On my own time I build independent tools that attack the gaps I kept running into: cameras that can't prove they were installed right, video that can't prove it's untampered, AI agents that can't prove what they did.

What I work on

• Verification. Signed video evidence and provenance (EdgeProof), and signed receipts for AI agent actions (AAR). If it can't survive a challenge, it isn't evidence.
• Commissioning. Tools that turn "the camera shows a picture" into a measurable pass or fail (camspec, the Commissioning Walk).
• Agent infrastructure. Making cameras and physical security systems something AI agents can actually operate (onvif-mcp).
• Writing. Notes from the Field, a series on what AI in physical security looks like when you've stood on both sides of it.