You ship production LLM features end-to-end — retrieval, prompting, tool-calling, evaluation and the boring-but-critical plumbing that makes them fast, cheap and reliable. You care about latency, cost and correctness in equal measure.
What you'll build
- Retrieval-augmented features that turn messy business documents into decisions.
- Tool-calling agents that safely operate inside our customers' workflows.
- Evaluation datasets and regression tests for every prompt in production.
- Instrumentation that lets us reason about cost, latency and quality per feature.
Requirements
- 4+ years software engineering; 1+ shipping LLM features to real users.
- Fluent in Python or TypeScript; comfortable across both.
- Understand embeddings, retrieval, prompt design and evaluation.
- Rigorous about testing, observability and cost.
- Right to work in the UK.
#J-18808-Ljbffr
