A portfolio is supposed to demonstrate depth. Mine is deliberately broad right now. Before someone clicks through and wonders why I've got half-a-dozen projects in half-a-dozen domains instead of one deep case study, it's worth explaining the choice.
The arc
I started writing code in the late '90s — PHP, MySQL, the kind of tooling where you learned the craft by wiring pieces together and watching what broke. Then I stepped away from development for a long stretch, working in operations where I saw a lot of software from the outside: the edges where tools meet real customers, the places things silently failed, the instinct for the plumbing between systems that the builders rarely notice.
I came back through AI. Specifically, through the moment where it became obvious that pairing with an LLM collapsed the cost of trying something new. A project that would have been a three-month ramp-up is now a three-day ramp-up. The rate-limiting step for modern software isn't code anymore — it's judgment about what to build.
Why I'm optimizing for surface area
I don't think I know yet what I should go deep on. That sounds like a liability; I think it's a feature of where I am. The more problems I've actually touched — not read about, not watched a talk on, but sat with long enough to ship a working version — the better the signal I have about which domains reward depth.
So right now I'm building across fairly different spaces: an MCP server aggregating real-world alert feeds, a monitoring tool for a failure mode I kept seeing in production, a local AI orchestrator because commercial assistants fall down at daily use, a document redaction pipeline because cloud services are the wrong answer for sensitive files. None of these share a stack or a customer. That's the point.
The tradeoff I'm making
Everything I ship is first-pass. No giant polished case studies. No shipped-at-Stripe logos. In exchange, I get to know what the shape of a problem looks like across many domains before I commit.
This isn't "jack of all trades, master of none." It's closer to: I want to be good at pattern recognition across trades. The interesting work in the next few years, I think, sits in the spaces between well-defined fields — and you can't see those spaces clearly from inside one of them.
What this site is
A working log, not a resume. The projects are real, live, and usually rough. The writing is what I've actually learned, not what I think sounds smart. If something here lines up with what you're working on, get in touch.