Loading video player...
In this episode, we create a full-stack Retrieval-Augmented Generation (RAG) application using PydanticAI, FastAPI, LanceDB, and Streamlit. We walk through each step, starting from setting up the backend with an agent, a vector database built with LanceDB, and an API layer using FastAPI. We demonstrate data ingestion from PDFs to text files, setting up vector embeddings, and running vector searches to retrieve relevant documents. Finally, we build a Streamlit frontend to interact with our AI model and showcase the complete stack. Stay tuned for the next episode, where we'll deploy the FastAPI application with Azure Functions. Github repo https://github.com/AIgineerAB/AI_engineering_four_weeks_course/tree/main/10_pydanticai_rags #rags #pydanticai #lancedb #fastapi 00:00 Introduction to Fullstack RAG Application 00:29 Understanding the Backend Components 03:31 Setting Up the Development Environment 08:04 Converting PDFs to Text 18:36 Ingesting Data into Vector Database 28:10 Implementing Time Delays and API Call Management 29:01 Setting Up and Running the Vector Database 30:09 Exploring the Knowledge Base 34:41 Building Data Models for Prompt and Response 36:42 Creating the RAG Model 44:37 Developing the API with FastAPI 47:56 Building the Streamlit Frontend 52:05 Testing and Debugging the Application 53:28 Conclusion and Next Steps