3-day RAG roadmap
Free3-day learning plan for Retrieval Augmented Generation
FreeFree tier
About 3-day RAG roadmap
The 3-Day RAG Roadmap is a free, open-source learning plan that takes learners from retrieval basics through building a real RAG app to advanced agentic techniques. Updated for 2026, it covers embeddings, vector search, chunking, reranking, hybrid search, evaluation, GraphRAG, and long-context tradeoffs. Each day requires about 2–3 hours and includes curated links to courses, papers, and research tables. The roadmap is part of the awesome-generative-ai-guide repository by aishwaryanr.
Key Features
Structured three-day curriculum (2–3 hours/day)
Day 1: Retrieval Foundations – embeddings, vector search, chunking
Day 2: Build a RAG App – chunking strategy, reranking, hybrid search, evaluation
Day 3: Agentic and Advanced RAG – agentic retrieval, GraphRAG, long-context tradeoffs
Curated links to DeepLearning.AI courses, foundational papers, and living research tables
Updated for 2026 with emphasis on agentic RAG
Pros & Cons
Pros
- Comprehensive three-day plan covering retrieval through advanced agentic RAG
- All resources are free and open source (hosted on GitHub)
- Includes links to reputable courses (DeepLearning.AI) and key research papers
- Updated for 2026, reflecting current trends like agentic control and GraphRAG
- Practical emphasis: from fundamentals to building an app with evaluation
Cons
- Requires self-motivation and dedicated 2–3 hours per day
- Assumes basic familiarity with large language models (LLMs) and Python
- Not a hands‑on coding tutorial; it is a curated set of resources and reading lists
- Some linked resources may become outdated; relies on external content maintained by others
Best For
Self-paced learning of RAG fundamentals and advanced techniquesBuilding a production‑ready RAG application from scratchUnderstanding agentic retrieval and when RAG is the right toolStructured study for developers and ML engineers new to RAG
FAQ
How much time does the 3-Day RAG Roadmap require?
The roadmap suggests about 2 to 3 hours per day for three days.
What topics are covered on Day 1?
Day 1 covers retrieval foundations: embeddings, vector search, and chunking.
Does the roadmap include practical building?
Yes, Day 2 moves from concept to a working pipeline including chunking strategy, reranking, hybrid search, and evaluation.
Is this roadmap suitable for beginners?
It starts with retrieval basics, but assumes basic familiarity with LLMs. Beginners may need additional background.
Is the roadmap free?
Yes, the roadmap is a free markdown file hosted on GitHub, and all linked resources are also free or open access.