About

I've been writing software since 2006, so this is my twentieth year. Whatever the technology, projects stalled in the same place. One person decided what to build and another wrote the code. The one who wrote the plan couldn't see the cost, and the one who wrote the code didn't know the reason. So now the same person does both. For companies that means AI features and existing code: I write down what the system should do, then build it, and I read the code you already have and say what to fix. For a small business I build a website, text included, and a custom web application when nothing off the shelf fits, from the first conversation to launch day.

By day I lead engineering at a large enterprise software company, and in the evenings I build and run my own products. Right now I'm testing in production how much of the coding can be handed to a machine — AI helpers that carry a task through on their own while I check what comes out. I run that test on my own products, where real traffic and revenue are at stake.

What that keeps teaching me is the distance between a demo and the real thing. An AI feature works beautifully in a demo, and rarely in a codebase that's been growing for years. The difference comes from the engineering around the model: how the data is stored, how long a user will wait for an answer, what happens when something fails. By that point the model and the instructions you give it are usually the smallest problem. The same holds for a website, with less at stake and a shorter deadline. That's the work I know, and that's what I take on.