About

Twenty years in software. Developer, architect, API advocate. And now, apparently, the person who asks awkward questions of corpora.

That last part wasn’t planned. It started with picking up a docs problem nobody assigned to me, which led to a question I hadn’t taken seriously before: what does documentation actually need to look like when AI agents are reading it, not humans?

Documentation debt doesn’t disappear when you adopt AI. It compounds.

Making the invisible visible — dashboards, gap analysis, cost checkers — that’s the actual work. Genuinely cross-disciplinary: I code enough to build pipelines, write enough to publish findings, and think architecturally enough to understand how knowledge fits together across a system.

What I keep finding: the problems that look like documentation problems aren’t. They’re about how engineering knowledge is structured, named, and organized, and whether humans and AI systems can actually use it. That’s made me a better reader of engineering knowledge — which turns out to be the harder skill.

I build experimental tooling to surface what’s invisible in engineering knowledge, for teams trying to make their systems work reliably with AI.

I write about what I find.

What that’s looked like lately

103 architecture standards restructured into an agent-readable corpus. 2,677 ADRs scraped and clustered to find a vocabulary too thin to search. A CLI I shipped on instinct, then rebuilt after 22 experiments told me my instincts were wrong about what actually matters. One free tool, markdown-ia, out in the world.

More on all of it on the Projects page.

How I like to work

If you’re wondering how I like to work, here’s the short version.

Get in Touch

You can find me on LinkedIn or reach out by email. I write about what I find on the blog.