How I Work —
From Idea to Production

We meet for thirty minutes on Zoom or by phone to understand what you want to build, the business pain behind the request, and your timeline. I'll give an honest read on whether the project is a fit — sometimes I'll refer you elsewhere if it isn't my domain. If there's a fit, I leave the call with a first-pass picture of scope, top risks, and a rough price range. If there isn't, we lost 30 minutes and nothing more.

This is where the idea becomes a working document. I write a PRD (Product Requirements Document) covering all features, user flows, edge cases, and integrations. In parallel I build mockups of the main screens (using tools like Stitch or v0) and propose a fitted technical architecture — which database, which services, how AI plugs in if relevant. By the end of this phase you have a 10–30 page document you can show to a board, a development team, or a partner — and I have a roadmap to build from.

The core of the work. I run weekly sprints — on Sunday we decide what goes into the round, on Thursday or Friday there's a demo of what shipped, and any feedback gets handled in the next round. You see real progress every week, not a month of silence followed by a big reveal. I work in Git with feature branches, clear commit messages, and documented PRs — so you have full visibility even when we're not in a meeting. Every commit passes type-check and lint, and I use AI tooling (Claude Code, Copilot) as a productivity multiplier — but the code itself goes through human review.

Before the system sees the light of day, we run a structured testing pass. Functional tests for every critical user flow, accessibility tests against IS 5568 / WCAG 2.1 (especially important for public-sector and enterprise projects), performance testing with Lighthouse and Core Web Vitals, and basic security checks (OWASP Top 10). Load testing if relevant. Every bug found goes into the list, gets fixed, and gets retested. This phase is short on time but critical to quality.

Launch day. I handle the full deployment — Vercel, Railway, a Hetzner VPS, or your internal cloud. I set up monitoring (Sentry for errors, Plausible/PostHog for analytics if relevant), confirm DNS, SSL, and the rest of the infrastructure are working, and run a final smoke test in production. After launch, you get 30 days of included support — every bug, every question, every small fix is handled at no extra charge. This is when the project really moves from "code" to "a live system".

After the included 30 days, if you want me to keep maintaining the system, I offer a flexible monthly retainer. The idea is straightforward: a guaranteed number of hours per month for bug fixes, security updates, small features, and fast incident response. You can also choose to take the code in-house and work with your own team — I'll run a clean handover and leave thorough documentation. No lock-in. My approach is to build systems that can run without me.