INTERMEDIATE
Build retrieval-augmented generation pipelines from scratch — document ingestion, chunking strategies, embedding models, vector databases, hybrid search, reranking, and evaluation frameworks for production RAG.
Why retrieval-augmented generation exists, how the retrieve-then-generate pipeline works, and the document ingestion and chunking decisions that determine everything downstream.
Embedding model selection, vector database architecture, HNSW indexing, metadata filtering, and the trade-offs between managed services and self-hosted vector stores.
Dense retrieval, sparse BM25, hybrid search with Reciprocal Rank Fusion, and cross-encoder reranking — how to find the right chunks for every query.
Context assembly, prompt construction for RAG, faithfulness and relevancy evaluation, RAGAS metrics, and advanced retrieval patterns like HyDE and query decomposition.
Running RAG in production — monitoring, feedback loops, re-indexing, multimodal retrieval, and a capstone challenge to build and evaluate a knowledge assistant for a real document corpus.
