I'm a software engineer who likes building small things that work. My projects include developer tools, machine learning inference on Apple Silicon, information retrieval, and automation that quietly does its job. This site is my working garden: a place for projects, technical notes, and the concepts I keep returning to while I build.
My bias is towards clarity, performance, and control. I'd rather understand a hundred lines completely than depend on a thousand I don't. Most of what I ship is minimal by default and auditable from source. That preference shows up in the shape of the site too. It is a set of static files, with the source public and the important writing available without a client-side application.
How I work
Most of what I know about reliable systems came from shipping software that had to survive timeouts, malformed input, interrupted deployments, and late-night crashes. I start simple, ship small changes, watch the path I touched, and keep a way back. A single server, a reverse proxy, and a short deployment script can go surprisingly far before the system asks for anything larger.
I write about the mechanisms behind the work: how state survives a process, why small diffs are easier to trust, where performance disappears, and what a tool changes in the person using it. The notes are first-person field records rather than a polished publication schedule. Some are seeds. Some have been revised enough to become evergreen.
Where to look
Browse the project index for open-source work, the writing garden for longer notes, and the word garden for smaller connected ideas. Source code and repository history live on GitHub. Machine-readable entry points are listed in llms.txt, the sitemap, and the developer resources page.