How I build

I like being involved in the whole product. The first conversation, the screens, the data, the awkward edge cases, and the moment it goes live.

AI is part of that process. I use Codex and Claude Code for implementation. Here is what the work around them looks like.

Get clear on the problem

I start by understanding who the product is for and what they need to do. I study references, map the roles and workflows, and choose the stack around those needs.

Give the build some shape

I break implementation into pieces I can follow and review. Each task needs clear behavior, constraints, and a way to tell when it is finished.

Open it and use it

I review the changes and test the actual experience. The screens need to make sense. The integrations need to work. The product needs to handle the cases outside the easiest path.

Get it out into the world

I handle hosting and deployment, check the live product, and keep refining it. A release is another chance to learn where the experience needs work.

That is the short version. There is more about my background here, and the tools I work with are on this page.