The short version
I lead a cloud-consulting practice for a global firm and I do it from here, with horses outside the window. Thirty-five years in the work, the last decade leading it, and I still write code most days. The best way I know to hold an opinion about a system is to run one.
What I care about most is making AI easy to adopt. Not a product — a stance. Anyone who touches it should see quickly what they can do, learn exactly how to do it, and come away with the thing itself, not a description of it. Everything I build, and most of what I write, comes back to that.
The work
I started as an engineer at Expedia and ended up running DevOps for CarRentals.com. Nordstrom made me a principal engineer. Then consulting: Contino, a cloud-native shop that was later folded into a much larger firm, where I've spent years owning the relationship with one of the world's largest financial institutions. For my engagements I am the sales team, the finance team, and the delivery team, which keeps me honest about what actually ships.
I don't have a degree. The thirty-five years are the credential.
What I'm good at: sitting with a CTO on strategy in the morning and writing the agent in the afternoon. Each half makes the other better; the strategy is sharper for having built the thing, and the build is cleaner for knowing why it matters.
What I'm building
Most of it is personal AI — agents that run real parts of a real household, on hardware I can see from my desk.
- AI-OS. One prompt that turns Claude Code into a personal AI operating system. Free, public, meant for someone who has never done this before. Start here.
- The ranch fleet. A few machines here run the solar and battery reporting, the water, the network, the weather station, the morning brief, and the checks that notice when any of it goes quiet. A local model floor keeps the important parts working when the internet or a provider doesn't. Corral is the one window I watch it all through.
- The principles. Every incident on the ranch gets turned into a rule the agents can reason from instead of a patch. Fifteen so far. Here's how that chain works.
I write about the failures more than the wins because the failures are where the rules come from.
The ranch
My wife Sheridan and I split ranch life between us: the horses, the dogs, the chickens, the cooking, and a hydroponic garden that fights the hard water every week. Two of the horses are mustangs; I wrote up the paperwork trail on where Red was born wild, which turned out to be a systems story too.
The place covers better than ninety percent of its own electricity from the solar on the roof and a battery in the garage. The office runs on it, the writing here runs on what I learn keeping it all alive.
If you grow food out here, I keep seventeen guides for this exact ground: hard water, cold nights, a winter window that runs backwards. And for my mother I traced the family back through four headwaters in Two Rivers.
Radio
I'm N0CVP. Roughly 6,900 contacts and a little over 200 DXCC entities, all from DM41 — you can walk the map if you like that sort of thing. The current radio project is a receive-only ionospheric observatory for the ranch, so I can see what propagation is doing instead of guessing.
How I work
- Hands on. I'd rather run the thing than review a slide about it, and I trust the person on the ground over the dashboard.
- I write decisions down, with the date, so I can argue with my past self instead of guessing what he was thinking.
- No single bet. The job, the product work, and the ranch each get their turn; when one starves, I notice.
- Sleep is sacred. Bed by nine, up by five, thirty pushups before I brush my teeth, and a mile or more on the ranch most days.
If any of this is useful to you, the notes are where it lands first, and every one of them ends with a place to tell me I'm wrong.