In this hands-on course, you will build a practical AI data assistant for real business data. The final project can answer questions from structured databases, NoSQL documents, business files, mock REST APIs, and a vector knowledge base. You will combine LLMs, Text-to-SQL, Retrieval-Augmented Generation (RAG), PostgreSQL pgvector, MongoDB query generation, tool-calling, routing, guardrails, testing, FastAPI, Streamlit, and Docker Compose into one end-to-end application.
The project uses a realistic enterprise scenario called BizData AI Assistant.
SQL Server stores CRM data such as customers, accounts, contacts, opportunities, activities, segments, and loyalty profiles. PostgreSQL stores commerce data such as users, products, inventory, orders, order items, payments, shipments, branches, and sales targets. MongoDB stores support tickets, product reviews, customer chats, and agent notes. PostgreSQL pgvector stores document embeddings for RAG over policies, SOPs, PDFs, Markdown files, and business documents. Local JSON files power a mock business API for shipment status, payment status, customer risk score, and inventory data.
