This is an MVP of a Q&A chatbot web application designed to answer questions using Dataiku document data. The chatbot service leverages a RAG (Retrieval-Augmented Generation) process and a Large Language Model (LLM), with Reactflow integrated to provide a clear visualization of the process flow. The llama-index framework was utilized to build the core chatbot functionality.
- Design and implement the user interface for the Chat page
- Design and build the database schema (PostgreSQL)
- Set up a vector store using pgvector for efficient similarity searches
- Develop the chat process from User Query to Output Response
- Add local LLM model(llama3, solar etc) serving via Ollama