Deciding
where the
seams go.

I'm Ikeji Joshua. I lead engineering at The Coast and I have been freelancing since December 2024. Most of my job is deciding where the seams go, then living with the consequences of that decision for the next year.

I work in TypeScript end to end. React and Next.js on the front, NestJS and Node behind it, MongoDB and Redis underneath, shipped in Docker. I am most useful on the parts that get expensive to fix later: auth and session revocation, subscription and payment state, anything where a wrong answer costs someone money or leaks their data.

I have run a few large migrations now, including moving a production app from one backend architecture to another without freezing feature work. Every time, the migration itself turns out to be the easy half. The contract between the old and new sides is where the real work is, and I have learned to write that down before touching code.

So I am heavy on written decisions: ADRs, architecture notes, an agreed vocabulary for the domain, so the same argument does not get relitigated every quarter. I am equally heavy on tests, including evals on the AI paths, because AI features fail quietly and nothing else catches it.

I build with AI tooling every day and I build AI features into products. I stay skeptical of both, which I think is the correct amount.