AI Engineer
Take AI features past the demo — evaluations, routing, fallbacks, and cost you can actually measure.
AI integrationRemoteFull-timeMid to senior
The prototype always works. This role exists because the gap between a convincing demo and a feature a customer pays for is almost entirely operational: what it costs per tenant, how it fails, and how you know quality dropped before a customer tells you.
You would own that gap — building the evaluation harnesses, the routing and fallback logic, and the metering that makes an AI feature a product rather than a science project.
At a glance
- Team
- AI integration
- Location
- Remote
- Type
- Full-time
- Level
- Mid to senior
What you would do
- Assess whether a use case is actually an AI problem, and say so when it is not
- Design model routing and fallback so one provider's bad day is not a customer's outage
- Build evaluation harnesses and quality gates that catch regressions before users do
- Instrument per-tenant usage and cost, and keep unit economics visible
- Own prompt and version management as a real engineering surface
What we are looking for
- You have shipped an LLM-backed feature to real users and maintained it afterwards
- Strong TypeScript or Python, and comfort working directly against model APIs
- You can design an evaluation for a task where correctness is not binary
- Clear-eyed about what current models can and cannot do reliably
Nice to have
- Retrieval systems and their failure modes
- Streaming interfaces and latency budgets
- Cost modelling across multiple providers
- Published writing or open source in the space
Apply
No formal cover letter needed. Tell us what you have shipped and what you would want to own here — we read every application and reply either way.