AI RAG Retrieval Model expert

**We will only hire someone who can prove they have solved this type of problem before. Please indicate this in your response, and the first meeting will be a demo of you solving this problem.**

We need someone who can update our existing AI RAG chatbot to answer questions that don't have "one" right answer. Here's a example of the types of questions we want to answer: Imagine you have a set of leases on PDFs that are chunked and embeddings stored in a vector db. Each lease has a different signatory, different dates, and different terms. We want to be able to ask questions like "How many leases do we have signed with *insert company here*?". Or "How many leases expire in the next thirty days?"

We aren't looking for one "right" answer, we want to chain our calls to survey the documents and research the right answer without running the query over every document.

We want to hire someone to improve our retrieval model to allow for this type of "survey and research" questions. Strong preference to someone who has solved this problem before, and we value a proven approach.