Engineering
AI Engineer
Decide where applied AI belongs in a product, and where it does not.
What you would do
- Design and ship AI features that hold up when the model misbehaves.
- Validate every model output before it reaches a person or a database.
- Keep token cost and latency inside a budget the product can afford.
What we are looking for
- Shipped an AI feature that survived contact with real users.
- Sceptical about model output by default, and build accordingly.
- Can explain why a feature should not use a model.