Chuyển đến nội dung chính
BlogCoursesHL7Cert PrepRoadmapGameDomainAboutMCPNEW
Topics
DatabaseDevOpsKiến trúc hệ thốngLinuxProgrammingSecurity

Real Battle RAG: From Basic to Advanced

In-depth course on Retrieval-Augmented Generation (RAG) — techniques for connecting the LLM to your own data. From basic RAG to Graph RAG, Agentic RAG, Multimodal RAG. Hands-on with ChromaDB, Qdrant, LangChain, LlamaIndex. Deploy "Chat with Documents" to production.

Introducing the Series

Retrieval-Augmented Generation is an in-depth course on Retrieval-Augmented Generation — a technique that allows LLM to access and use your private data (internal documents, knowledge base, database) to answer accurately and up-to-date.

🎯 Why is RAG important? RAG is the #1 technique applied by businesses to solve the 3 biggest problems of LLM: hallucination, knowledge cutoff (old knowledge), and domain-specific knowledge (specialized knowledge).

What will you learn?

Part 1: RAG Platform

  • Lesson 1: What is RAG? Architecture Retrieve → Augment → Generate
  • Lesson 2: Embedding Models: OpenAI, Cohere, Open-source for Vietnamese
  • Lesson 3: Vector Databases: Chroma, Qdrant, Pinecone — setup & comparison

Part 2: Document Processing Pipeline

  • Lesson 4: Document Loading: PDF, DOCX, Web, YouTube, Code repos
  • Lesson 5: Chunking Strategies: fixed vs semantic vs recursive
  • Lesson 6: Metadata, Filtering & Hybrid Search

Part 3: Advanced RAG Techniques

  • Lesson 7: Query Transformation: HyDE, Multi-Query, Step-Back
  • Lesson 8: Re-Ranking & Contextual Compression
  • Lesson 9: 🔥 Graph RAG — Knowledge Graph + Vector Search
  • Lesson 10: 🔥 Multimodal RAG — Photos, tables, charts in documents

Part 4: Production RAG

  • Lesson 11: 🔥 Agentic RAG — Agent decides when to retrieve
  • Lesson 12: RAG Evaluation: RAGAS framework
  • Lesson 13: Deploy to Production: API, caching, monitoring
  • Lesson 14: Capstone: "Chat with Documents" complete

Input required

  • Intermediate Python (async/await, file I/O, API calls)
  • Basic understanding of LLM and Prompt Engineering
  • OpenAI or Anthropic account (for embedding + LLM calls)

Tools used

Python 3.11+        | Ngôn ngữ chính
OpenAI / Anthropic   | LLM APIs + Embeddings
ChromaDB / Qdrant    | Vector Databases
LangChain            | RAG framework
LlamaIndex           | Alternative RAG framework
Unstructured.io      | Document processing
Neo4j                | Graph database (Graph RAG)
RAGAS                | RAG evaluation
FastAPI              | Production API

Part 1: RAG Platform

Lesson 1: What is RAG? — Architecture, Use Cases & Why RAG is needed90 minLesson 2: Embedding Models — Turn Text into Vector120 minLesson 3: Vector Databases — Chroma, Qdrant, Pinecone, Weaviate150 min

Part 2: Document Processing Pipeline

Lesson 4: Document Loading — PDF, DOCX, Web, YouTube, Code120 minLesson 5: Chunking Strategies — Fixed, Semantic, Recursive150 minLesson 6: Metadata, Filtering & Hybrid Search150 min

Part 3: Advanced Query & Retrieval

Lesson 7: Query Transformation — HyDE, Multi-Query, Step-Back150 minLesson 8: Re-Ranking & Contextual Compression150 minLesson 9: Graph RAG — Knowledge Graph + Vector Search150 min

Part 4: Advanced RAG Patterns

Lesson 10: Multimodal RAG — Images, Tables, Charts in Documents150 minLesson 11: Agentic RAG — Agent + RAG combines power150 minLesson 12: RAG Evaluation — RAGAS, Faithfulness & Relevancy150 min

Part 5: Production & Capstone

Lesson 13: Deploy RAG to Production — API, Caching & Monitoring150 minLesson 14: Capstone — Build a complete Chat with Documents150 min