RAG Engineer Needed for Retrieval + Citation Enhancement in Existing AI Knowledge Assistant

We have an existing AI knowledge assistant that answers questions over company documents using a RAG pipeline. The core app already exists. I need an experienced RAG / LLM developer to improve one focused part of the system: retrieval quality and citation grounding. This is a small enhancement task inside a larger project, not a full chatbot build. *Use Case:* The assistant answers questions over internal compliance / policy documents. We want answers to be better grounded in retrieved document chunks, with clearer source citations and better handling when the answer is not present. *Scope:* •⁠ ⁠Review the existing RAG flow •⁠ ⁠Improve retrieval prompt / logic for policy-style questions •⁠ ⁠Improve chunk selection before answer generation •⁠ ⁠Add or improve citation formatting in final answers •⁠ ⁠Add “not found in documents” behavior when context is insufficient •⁠ ⁠Test with 5–6 sample policy questions *Example Questions:* •⁠ ⁠“Who approves vendor payments above $10,000?” •⁠ ⁠“What documents are required for employee onboarding?” •⁠ ⁠“What is the reimbursement limit for travel?” •⁠ ⁠“What should the assistant say if the policy does not mention this?” *Expected Deliverable:* •⁠ ⁠Small PR or committed code changes •⁠ ⁠5–6 tested example queries •⁠ ⁠Brief notes on what was improved •⁠ ⁠Clear explanation of any remaining limitations *Preferred Experience:* •⁠ ⁠RAG systems •⁠ ⁠Python or TypeScript •⁠ ⁠LangChain, LlamaIndex, or custom RAG pipelines •⁠ ⁠Vector search / embeddings •⁠ ⁠Prompt engineering •⁠ ⁠Citation-grounded answer generation •⁠ ⁠Reducing hallucinations in LLM apps *To Apply, Please Answer:* 1.⁠ ⁠Have you improved an existing RAG system before? 2.⁠ ⁠How would you make answers more grounded in retrieved context?