Azercell Policy Assistant
Multi-session RAG chatbot over corporate policy and internal docs.
The Azercell Policy Assistant is a multi-session retrieval-augmented generation (RAG) chatbot that lets employees ask natural-language questions about corporate policy, ethics guidelines, and internal documentation — and get grounded answers instead of a search-results dump.
It supports multi-session conversations, so context carries across a dialogue rather than resetting on every message. The backend is a FastAPI service handling vector-based retrieval and orchestration against AWS Bedrock models, with a Streamlit frontend for the chat UI.
The whole system ships as a Dockerized full-stack application with CI/CD automation via GitHub Actions, so changes are built, tested, and deployed reproducibly rather than by hand.