I started in graphic design, caring mostly about type and visual hierarchy. Around 2021 I realized design without code was half the job, so I moved into engineering.

Today I build full-stack systems end to end. Frontend, backend, data, AI, payments, deployment. I'm comfortable across the whole surface.

I'm most useful when the problem cuts across parts of a product and someone has to hold the whole thing in their head until it works. I'd rather ship a working version in a week than a perfect spec in a month.

I try to build things that bend when the business changes direction, because the business always does. That means strict typing, clear module boundaries, and not being clever when boring would work. Rigid systems are fast to write and painful to keep. Bendy systems cost more up front and a lot less to live with.

I work in short cycles with running demos, not Figma files. Architecture decisions show up in production so we can see whether they hold. Technical debt gets written down out loud, not hidden for the next engineer to find.

The job isn't to ship code. It's to make sure the codebase six months from now is still somewhere the next engineer wants to work.

AI is my compiler, not my architect.

I use coding agents for scaffolding, refactors, test generation, and reading new codebases. They're fast where fast is useful. They don't pick the architecture, validate business logic, or sign off on quality. Those stay with me.

Anything that touches money, eligibility, user-visible correctness, or compliance gets tests, evals, and a manual review before it ships. I keep a trail of where AI helped and where I overrode it. That makes things much easier to debug six months later when something goes sideways and someone needs to find the root cause fast.

Yes, I ship faster with AI. No, I don't ship AI output I can't explain.