
Already know some of this? A 2-minute test skips the lessons you know.
Take the placement testA free 23-lesson course built from DSwithBappy's 3h 44m of YouTube video. 3 days at 90 minutes a day, with study notes and a quiz in every lesson. A 2-minute placement test skips what you already know.
Taught by DSwithBappy on YouTube. Videos play in YouTube's own player, so the creator keeps every view.
Learn the agentic RAG workflow: private knowledge base first, evidence grading, web search fallback, query rewriting, and grounded answer generation with citations.
Review the production-ready stack (LangGraph, Pinecone, Tavily, FastAPI, DigitalOcean) and initialize the GitHub repository with folder structure.
Create a virtual environment, pin dependencies in requirements.txt, and build a centralized Pydantic settings class with structured logging.
Implement a modular data injection service that loads PDF, TXT, MD, and DOCX files, then splits them into overlapping chunks using LangChain's recursive character text splitter.
Load a Hugging Face sentence-transformer embedding model, create a Pinecone serverless index with namespaces, and upsert chunk vectors with metadata for retrieval.
Define the shared LangGraph state including question, current query, retrieved documents, grades, retry count, trace, and citation fields.
Build the full LangGraph workflow: router, KB retrieval, evidence grading, Tavily web search, query rewriting, answer generation, and edge connections with retry loop termination.
Expose health check, chat, and document upload endpoints via FastAPI, integrating the agentic workflow and vector store ingestion.
Develop a minimal HTML/CSS/JS single-page chat UI that calls the FastAPI backend and displays answers with execution traces and citations.
Run the full stack locally, test private KB answers, web fallback, direct LLM responses, and dynamic document ingestion into Pinecone.
Write a multi-stage Dockerfile, configure DigitalOcean App Platform with GitHub CI/CD, inject environment variables, and deploy the production-ready service.