Back to Careers
RemoteProject-based / Advisory

RAG & Retrieval Specialist

Embeddings, chunking, hybrid search, reranking. Fix wrong answers and low recall with measurable benchmarks.

About the role

We need a RAG specialist who can diagnose and fix retrieval failures. Our clients have chatbots giving wrong answers despite good docs—often it's retrieval, not the model. You'll improve chunking, embeddings, hybrid search, reranking, and prove it with golden sets.

Responsibilities

  • Diagnose retrieval failure modes—low recall, low precision, irrelevant context
  • Optimize chunking strategies, embedding alignment, and reranking pipelines
  • Build golden sets and eval harnesses for retrieval quality
  • Ship fixes with before/after benchmarks on retrieval hit rate and answer quality

Requirements

  • Deep experience with RAG systems—embeddings, vector DBs, chunking, reranking
  • Comfortable with Python and common RAG tooling (LangChain, LlamaIndex, or similar)
  • Evidence-based—measure retrieval quality, not guess
  • Strong problem-solving for production systems

Nice to have

  • Experience with hybrid search, metadata filters, query rewriting
  • Familiarity with citation and groundedness evaluation

What we offer

  • Remote-first, async-friendly
  • Learn from real production failures—the best teacher
  • Flexible engagement—project-based, no lock-in

Ready to apply?

No formal application. Send us a short intro, relevant work, and how you'd like to collaborate.