I make things, mostly software,
and write down what I learned.
I grew up in Queens, New York, middle child, and started running a webstore called odeshop.com when I was a teenager because the obvious thing to do with the internet, when you discovered you could build on it, was to build something on it. It failed, but the store didn't matter. The making mattered.
That instinct never left. Twenty years later it's the same instinct, with better tools and harder problems.
Engineering became the formal version of it. I worked at a financial firm, then within the web of many company networks where I touched everything from the rack to the application, then a stretch at a large bank doing platform work that taught me what regulated software actually requires. Each step taught me what the previous one was missing. The arc was always toward more leverage — doing more with the same hands.
The current chapter started one fateful day when my brother Chris showed me what I'd been missing with AI. I decided then, that the next step in leverage wasn't another framework or another platform. It was turning the practice itself over to AI agents, and learning to operate the workforce instead of writing the code. This site is the field notes from that.
Outside of work I dedicate myself to being good at the three Ds. Dad. Dude. husbanD. The most interesting people I know are the ones who don't let the world weigh them down. I try to be one of those.
I do my best work with the right surface around me — the platform, the agents, the team. It's more about the tools I have at my disposal and how easy it is to reach them; be it people, an AI agent, or a dev server. I produce fast loops and clean attribution when focused with the right tools at hand.
I'm allergic to ceremony that doesn't pay. I'm protective of ceremony that does. The skill is knowing which is which, and naming it explicitly so the next operator (often future-me) doesn't have to relearn the distinction.
Quiet by default. Direct in conversation. Slow to commit to a shape, fast to retire one once it's wrong.
For the operator-side detail — the structured assessments the AI agent writes at the close of each milestone, sanitized for public reading — see operator assessments.
For what I'm focused on now, see /now.
For the toolkit I actually use day-to-day, see /uses.