Description: We are looking for an experienced AI engineer to help us build a retrieval-augmented-generation (RAG) assistant using sensitive internal knowledge. This is a sourcing / candidate search — we are currently exploring talent and not yet ready to deploy. The role (for sourcing purposes) We want to understand who has experience in: Building LangChain- or LlamaIndex-driven pipelines that index internal documents from scratch. Integrating open-source LLMs such as Llama 3 and Mixtral in a secure, on-premises environment. Designing data ingestion and embedding strategies for PDFs, slides, markdown, or other internal documents. Creating RAG pipelines with retrievers, optional re-rankers, and generation stages. Optimising for fast, accurate retrieval and maintaining strict data confidentiality. Key skills / experience we are looking for: Python development, PyTorch, Hugging Face Transformers, GPTQ or vLLM experience. Previous delivery of RAG chatbots or similar knowledge-base assistants. Ability to work with sensitive data and design pipelines that run fully offline. Familiarity with vector databases such as Qdrant, Chroma, or Pinecone. NDA / Confidentiality: All applicants must be willing to sign an NDA before discussing project details or receiving any sample data. Next Steps: This posting is for candidate sourcing only — we’re collecting qualified engineers for a confidential AI project and will follow up with selected candidates when ready to move forward. Budget / Availability: TBD (to be discussed with shortlisted candidates).