A working product, not a prototype — web, mobile, or both — with AI built in where it earns its place.

Most builds die on the parts nobody scopes: authentication that survives contact with real accounts, a data model you won't have to unpick in month three, and a deploy pipeline so shipping isn't an event. I build those first, then the feature you actually pitched. When the product calls for a model — reading documents, triaging intake, flagging anomalies — it goes in with a review queue, guardrails and an audit log, so it stays switched on after launch.

Web MVP (from $4,000, ~6 weeks). One core flow finished end to end, on real infrastructure, with auth and a sound data model. Next.js, Node, PostgreSQL, deployed to your cloud or mine.

Dashboards and admin panels. Role-based access, reporting, and the internal tools your operations team actually uses. I have shipped permission systems covering 300+ users and 1,000+ distinct actions.

Mobile apps (React Native / Expo). iOS and Android from one codebase. Glean, my on-device iPhone document scanner, is live on the App Store — designed and built end to end.

AI inside the product. Agents and model features wired into your data — deterministic where determinism matters, the model only where judgement is genuinely needed.

Ongoing engineering ($65/hr). Features, refactors, integrations and stabilising a codebase you inherited, once the product is live and you are reacting to real users.

Model the domain. Before any UI, we settle the data model and who is allowed to do what. Getting this wrong is the single most expensive mistake in an MVP.

One core flow, finished. Rather than six half-features, one path a user can complete end to end.

Ship to a real environment. Deployed, on your domain, with the pipeline in place so the next change takes minutes.

Hand over honestly. You get the repo, the architecture, and a straight account of what is production-ready and what is deliberately deferred.

I've built multi-tenant SaaS with role-based permissions across customers, suppliers and internal staff; a payroll and compliance platform that has run monthly since May 2026; and AI agents that handle order intake in production. That is the experience I bring to scoping what your product genuinely needs versus what can wait.