📚 RAG & Vector Search — Topic Resources
Curated list of top landmark papers, open-source repositories, video series, free courses, and code references for Retrieval Augmented Generation (RAG), Hybrid Search, Vector Databases, Reranking, and Context Engineering .
📄 Landmark Papers & Essential Reading
Paper / Reference
Key Takeaways & Focus
Link
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (Lewis et al., 2020) Foundational paper combining parametric memory (generator) with non-parametric memory (retriever).
ArXiv Link
Dense Passage Retrieval (DPR) (Karpukhin et al., 2020) Proved dual-encoder dense vector retrieval outperforms traditional sparse BM25 for open-domain QA.
ArXiv Link
ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction (Khattab et al., 2020) Late-interaction model preserving fine-grained token-level matching with sub-second vector search speed.
ArXiv Link
Precise Zero-Shot Dense Retrieval without Relevance Labels (HyDE) (Gao et al., 2022) Hypothetical Document Embeddings — generating candidate answers first to retrieve relevant real documents.
ArXiv Link
From Local to Global Retrieval-Augmented Generation (GraphRAG) (Edge et al., Microsoft 2024) Combines knowledge graph extraction with LLM community summaries for global dataset understanding.
ArXiv Link
Corrective Retrieval-Augmented Generation (CRAG) (Yan et al., 2024) Evaluates document relevance dynamically and triggers web search fallbacks for low-confidence retrievals.
ArXiv Link
Speculative RAG: Enhancing Retrieval-Augmented Generation via Draft Models (Wang et al., 2024) Uses a small specialist draft model to evaluate multiple document subsets in parallel for fast verification.
ArXiv Link
LightRAG: Simple and Fast Knowledge Graph RAG (Guo et al., 2024) Dual-level retrieval paradigm combining entity-relation graphs with low-overhead vector search.
ArXiv Link
💻 Top Open-Source Repositories & Vector DBs
Repository
Description
Link
LlamaIndex The premier data framework for connecting private data sources (PDFs, DBs, Slack) to LLMs with advanced RAG.
GitHub Repo
LangChain Popular framework for building context-aware reasoning applications and retrieval pipelines.
GitHub Repo
Qdrant High-performance vector database written in Rust with vector payload filtering and hybrid BM25 search.
GitHub Repo
Chroma Open-source embedding database designed for easy local development and Python/TypeScript integration.
GitHub Repo
Milvus Enterprise-grade cloud-native vector database built for billion-scale vector retrieval and distributed cluster setup.
GitHub Repo
pgvector Open-source vector similarity search extension for PostgreSQL supporting HNSW and IVFFlat indexes.
GitHub Repo
LanceDB Developer-friendly, serverless vector database built on Apache Arrow for fast columnar and multimodal search.
GitHub Repo
🎥 Must-Watch YouTube Videos & Free Lectures
Video / Playlist
Creator
Description
Link
Building Production RAG Applications DeepLearning.AI
Free short course by Jerry Liu (LlamaIndex founder) on advanced RAG patterns (sentence window, auto-merging).
DeepLearning.AI Site
RAG Architecture & Vector Search Masterclass James Briggs
Step-by-step video tutorials on Pinecone, Qdrant, hybrid search, Cohere reranking, and BM25 integration.
YouTube Channel
RAG Triad & Evaluation TruLens / Arize
Video walkthroughs explaining Context Relevance, Groundedness, and Answer Relevance evaluation.
YouTube Video
🎓 Free Courses & Open Curricula
Course Title
Institution / Host
Focus
Link
Advanced RAG with LlamaIndex DeepLearning.AI
Practical hands-on training on document parsing, indexing, query routers, and reranking algorithms.
DeepLearning.AI Course
Vector Search & Embeddings Academy Pinecone / Weaviate
Interactive guides covering vector distance metrics (Cosine, Dot Product, Euclidean), HNSW, and IVF index tuning.
Weaviate Academy
RAG Techniques (Open Source Hub) Nirant Kasliwal
Comprehensive directory of RAG strategies including Parent-Document, GraphRAG, and Multi-Vector retrieval.
GitHub Repo
⚙️ Code References & Hands-on Notebooks