Build a RAG Chatbot for Our Internal Documents

We’re looking for an experienced AI/RAG developer to build a simple chatbot that can answer questions based strictly on our collection of PDFs and internal documents. The main priority is answer accuracy, reliable document retrieval, and preventing hallucinations. We don’t need a complex or highly designed user interface — we want a solid RAG pipeline that can retrieve the right information and provide trustworthy answers from our knowledge base. What We Need: 1. Ingest and process our PDF/internal documents 2. Split and embed documents appropriately 3. Store and retrieve relevant content using a vector database 4. Use an LLM such as OpenAI or Claude 5. Implement a reliable Retrieval-Augmented Generation (RAG) pipeline 6. Answer questions based only on information available in the knowledge base 7. Avoid making up information when the answer cannot be found 8. Clearly indicate when the available documents do not contain an answer 9. Provide relevant source/document references where practical 10. Handle follow-up questions and normal conversational queries Preferred Technology: Experience with the following is strongly preferred: 1. RAG / Retrieval-Augmented Generation 2. OpenAI API and/or Claude API 3. LangChain or similar frameworks 4. Vector databases such as Pinecone, Chroma, Qdrant, Weaviate, or FAISS 5. Python PDF/document processing 6. Embeddings and semantic search 7. Prompt engineering and hallucination reduction You don’t necessarily need to use every technology listed above. We’re open to your recommendations if you can explain why a different approach would provide better results. Key Requirement: Accuracy This project is primarily about retrieval quality and trustworthy answers, not UI design. The chatbot should not invent information. If the relevant information cannot be found in the documents, it should respond appropriately rather than guessing. We would also like the implementation to be structured so that the RAG pipeline can be tested and improved as we add more documents. Deliverables: 1. Working RAG chatbot 2. Document ingestion pipeline 3. Vector database setup 4. LLM integration 5. Retrieval and response-generation pipeline 6. Basic chatbot interface/API 7. Source/reference information for retrieved answers, where applicable 8. Basic instructions for running and updating the system 9. Clean, maintainable code Please apply if you have hands-on experience building RAG applications, especially document-based chatbots. When applying, please include: 1. Examples of similar RAG projects you have built 2. Which LLMs and vector databases you have worked with 3. Your preferred RAG stack for this project and why 4. How you approach reducing hallucinations 5. How you evaluate whether retrieved documents are relevant 6. Your estimated timeline for completing the project Please do not send a generic AI/ChatGPT proposal. We are specifically looking for someone who has actually built and deployed RAG/document-question-answering systems.