What we’re looking for

An engineer who can own features end to end in Next.js, TypeScript and Supabase, and who stopped treating AI as autocomplete a long time ago.

Two things matter here in equal measure.

First, real engineering foundations. Architecture, systems thinking, clean TypeScript, sane data modelling and the reliability instincts that only come from shipping to actual users. You don’t need hand-holding on trade-offs. You’ve made enough bad calls to know why the boring decision is usually the right one.

Second, you build the agents rather than just using the tools. Code review, bug triage, test generation, the routine maintenance nobody volunteers for. And you have a clear view, one that keeps changing as the tools do, on where AI genuinely speeds delivery up and where it creates a bigger mess for someone else to clean up later.

This isn’t a role for someone who treats AI as a shortcut. We want a pragmatic builder who has properly changed how they work: faster, thinking at a higher level and shipping more than was possible before. You’ll own slices of architecture, shape technical direction and raise the bar for the juniors around you. No staff-level title required.

What you’ll be responsible for

• You’ll build and maintain the core product flows in Next.js (TypeScript, app router, server components, API routes) that real customers use every day.

• You’ll design and evolve our Supabase schemas, RLS policies and Edge Functions, and keep the data models clean and documented as the product grows.

• You’ll keep the data flow between frontend and Supabase honest: types, queries, caching and error handling.

• You’ll use Cursor, Claude Code or whatever you prefer daily to scaffold features, write tests, refactor and evolve schemas, while still making the architectural calls yourself.

• You’ll build, configure and tune AI coding agents that run across the development lifecycle. Agents that triage bug reports, suggest fixes, enforce standards on pull requests or run targeted regression tests on the modules that changed.

• You’ll review human and AI-written code and keep AI spaghetti out of the codebase. You’re the quality gate, not the AI.

• You’ll ship from idea to deployed to iterated, with a bias to owning the whole slice rather than handing it on halfway.

• You’ll keep an eye on new AI development tools and agent frameworks, and tell us what’s worth adopting and what’s noise.

• What you won’t own: prioritisation and the client relationship. What gets built and in what order sits with delivery and the founders, though you’ll often be the person telling us what something really costs.

A bit more about you

• You’ve spent a few years in web engineering, at least a couple of them in modern TypeScript-heavy work.

• You’re strong with Next.js and React: routing, server components, data fetching, API routes and the performance basics.

• You’ve worked with Supabase or a similar Postgres platform. Schema design, Auth, RLS, migrations and enough performance tuning to stop things falling over at the worst possible moment.

• You’re a power user of AI coding tools. Not side experiments, not a weekend thing. Shipping with AI is just how you build.

• You’ve built or configured agents for engineering workflows: code review automation, bug triage, test generation pipelines or similar. Show us what you built.

• You’ve integrated AI APIs (OpenAI, Anthropic and the rest) into production apps.

• You work well on your own in a startup. Take a rough problem, shape a solution, get it live.

• You’re happy wearing lots of hats. We’re small and this spec will be out of date within a quarter.

Nice to have

• Supabase Edge Functions, real-time and the more advanced Postgres features (views, triggers, background jobs)

• Tech lead experience on small projects: folder structure, library choices, setting the conventions everyone else then follows

• CI/CD, testing infrastructure and cloud infrastructure exposure as a product scales

• Something you’ve built and put in front of real users, side project or otherwise

How this process works

1. Intro call, 30 minutes

2. Technical conversation with our engineering team. Walk us through something you owned end to end, including the parts that broke.

3. Practical exercise: paired build session, short take-home or a walkthrough of an agent you’ve built

4. Final conversation with Dan and Matt

What we’re offering

Real ownership of architecture on a product clients actually use, not a demo that lives in a deck.

Somewhere you can use AI properly, with the budget and the tools to back it up. It’s the job, not a side project.

A permanent, fully remote role and the potential to participate in our share option scheme.

Direct access to the founders and to our clients. No layers.

Ready to build with us?

If this sounds like you, we’d love to hear from you. Send us your CV and tell us what you’ve built.