We run an AI-native software team: contract-first, Claude at every step, features live behind a flag in days - no sprints, no staging, no hand-offs. We're hiring two full-stack engineers (Python + JS/TS) who'll master that pipeline on our product, then deploy forward to execute external projects on it.
Both roles are full-stack by default - backend, web, and app in the same week - and both grow into forward-deployed engineers who run entire external projects on the loop.
0–2 years. Solid Python + JS/TS fundamentals and genuine AI-native fluency. Ship well-scoped features from your first weeks, grow into the Engineer DRI seat, then deploy forward. Slope over pedigree - new grads welcome.
View role →5+ years. Deep Python + JS/TS. Own the biggest features end-to-end from day one, hold the technical bar at the review gates, mentor the junior engineer, and lead forward-deployed engagements.
View role →Every feature runs a 10-stage pipeline where Claude does the heavy lifting and a human DRI - one directly responsible engineer - drives every step.
A frozen backend↔frontend contract + simulator before any code - so both sides build in parallel against the same reality.
Claude Code writes most of the code. Engineers direct it, review it, and verify it by running the app - not by trusting the diff.
Every screen is composed from a living component library; anything new is contributed back. Build it in the system, not the feature.
Verify locally, merge dark behind a feature flag in production, dogfood there, flip it on. Rollback is one click.
Each step outputs a structured doc the next step's Claude reads directly. Three synchronous touchpoints per feature - that's it.
Screenshots, green gates, a flag, and a one-click rollback - never just "code written."
A forward-deployed engineer ships working software while embedded with an external team - owning the project from the scoping conversation to the live feature, on their stack, at our speed.
The centrepiece is a paid, timeboxed build exercise where using AI isn't just allowed - it's required. We assess how you direct and verify AI, not typing speed.
Apply with links to things you've shipped - working software beats résumé lines. Location and compensation are discussed during the process.