AI RAG Engineer (City of London)

We're a fast-moving startup building intelligent systems powered by large language models. We're looking for a software engineer who can design, build, and deploy AI systems end-to-end — especially those involving retrieval-augmented generation (RAG) , embeddings, and LLM orchestration. This is a hands-on role with real impact and ownership.

What You'll Work On

• Architect and implement RAG pipelines that combine vector search, custom retrievers, and LLM reasoning
• Own the evaluation stack — design eval harnesses, benchmarks, and regression tests for LLM outputs
• Build and scale infrastructure for deploying models and agents into real production environments
• Experiment with model behavior, latency trade-offs, and prompt tuning
• Collaborate closely with founders on product, architecture, and research priorities

✅ Core Requirements

• Proven experience building AI systems with LLMs — you’ve worked with tools like LangChain, LlamaIndex, Haystack, or built your own stack
• Hands-on with embedding models, vector DBs (e.g., FAISS, Weaviate, Qdrant), and retrieval logic
• Strong Python engineering skills — you write clean, production-ready code with tests
• Experience building and evaluating RAG pipelines in a real-world setting
• Familiarity with LLM evaluation techniques — you don’t deploy until you’ve tested against real metrics
• Solid understanding of modern cloud infrastructure (e.g., Docker, Kubernetes, serverless, GCP/AWS)

Bonus Skills

• Built custom eval pipelines using tools like Ragas , TruLens , or your own scoring systems
• Experience tuning open-source models (e.g., Mistral, LLaMA, Falcon) or working with APIs (OpenAI, Anthropic, Cohere)
• Exposure to agentic systems , tools + memory management, or multi-step reasoning chains
• Experience in fast-paced, early-stage startup environments

Why This Role Is Unique

• You’ll be engineering AI features that ship to users , not just running experiments
• Evaluation is first-class — we’re serious about quality, not just it looks good in the demo
• You’ll help shape both tech strategy and engineering culture from day one
• We care more about what you’ve built than where you’ve worked

You should apply if:

You're an engineer who enjoys building production-grade AI systems , and you believe evals, not vibes, should drive development. You're comfortable moving fast, debugging strange model behavior, and taking real ownership of the tech.

Note, this company is sponsoring visas