30 Days of RAG (Tutorial Series)

Date: May 1, 2026
Maintainer: Hassan Khan Email: hassanaiengineer@gmail.com
LinkedIn: https://www.linkedin.com/in/hassan-khan-4961b722b/

This repository is a hands-on 30-day RAG series: small projects (FastAPI + Streamlit) covering retrieval patterns, routing, hybrid search, knowledge graphs, multimodal (vision) RAG, and debugging/failure diagnostics.

Motivation and goals: see MOTIVATION.md.

Tech Stack (Across the Series)

Not every tutorial uses every tool, but you’ll see many of these across days:

• FastAPI: https://fastapi.tiangolo.com/
• Uvicorn: https://www.uvicorn.org/
• Streamlit: https://streamlit.io/
• Sentence Transformers: https://www.sbert.net/
• FAISS: https://github.com/facebookresearch/faiss
• Qdrant: https://qdrant.tech/
• Neo4j: https://neo4j.com/
• Ollama (local LLMs): https://ollama.com/

Quick Start / How to Run

Each numbered folder is an independent, self-contained project.

For API Tutorials (Days 01–08)

These are built with FastAPI. To run them:

cd 01_simple_rag_api
pip install -r requirements.txt
python -m app.ingest
uvicorn api.main:app --reload

For UI Tutorials (Days 09–30)

These are built with Streamlit. To run them:

cd 10_hybrid_search_rag_raglite
pip install -r requirements.txt
streamlit run main.py

Note: Some tutorials require API keys (OpenAI, Gemini, etc.) or external services (Qdrant, Neo4j). Use the .env.example file in the root directory as a starting point. If your system defaults to Python 2, make sure to use python3 and pip3.

Tutorials

🟦 Phase 1: Foundation (Days 01–08)

Focus: Core RAG patterns using FastAPI, local embeddings, and offline logic.

Day	Folder	Type	Topic	Key Tech
01	01_simple_rag_api/	FastAPI	Minimal RAG API	FAISS + SentenceTransformers
02	02_multiquery_rag_api/	FastAPI	Multi-query expansion	Offline Recall Optimization
03	03_self_rag_api/	FastAPI	Self-RAG loop	Iterative Refinement
04	04_tool_augmented_rag_api/	FastAPI	Tool-augmented RAG	Agentic Tool Scaffolding
05	05_corrective_rag_api/	FastAPI	Corrective flow (CRAG)	Retrieve → Evaluate → Refine
06	06_hybrid_rag_api/	FastAPI	Hybrid retrieval	Semantic + BM25 Fusion
07	07_graph_rag_api/	FastAPI	Graph-style expansion	NetworkX Knowledge Paths
08	08_agentic_rag_api/	FastAPI	Agentic orchestration	Autonomous API Logic

---

🟪 Phase 2: Cloud & UI (Days 09–16)

Focus: Streamlit dashboards, PDF chat, and integrating major LLM providers (Google, OpenAI).

Day	Folder	Type	Topic	Key Tech
09	09_rag_chain_pdf_chat/	Streamlit	PDF Document Chat	Gemini + ChromaDB
10	10_hybrid_search_rag_raglite/	Streamlit	Hybrid search + Reranking	RAGLite + Cohere
11	11_local_hybrid_search_rag_raglite/	Streamlit	Local hybrid with GGUF	Llama-cpp + UI
12	12_local_rag_agent_agno_ollama/	Python	Local RAG agent	Qdrant + Agno
13	13_llama_webpage_rag/	Streamlit	Webpage chat	Ollama + Chroma
14	14_agentic_rag_gpt5_agno/	Streamlit	URL Context Indexing	LanceDB + OpenAI
15	15_agentic_rag_embedding_gemma/	Streamlit	Local retrieval agent	EmbeddingGemma + LanceDB
16	16_agentic_rag_with_reasoning/	Streamlit	Reasoning-based RAG	OpenAI + Gemini

---

🟧 Phase 3: Local LLMs (Days 17–20)

Focus: Running cutting-edge local models (DeepSeek, Qwen) for specific RAG tasks.

Day	Folder	Type	Topic	Key Tech
17	17_agentic_rag_math_agent/	Streamlit	Math & Logic Agent	SymPy + Ollama
18	18_gemini_agentic_rag/	Streamlit	Production Qdrant RAG	Gemini + Qdrant Cloud
19	19_deepseek_local_rag_agent/	Streamlit	DeepSeek reasoning	Ollama + Qdrant
20	20_qwen_local_rag_agent/	Streamlit	Model-routing UI	Qwen/Gemma Selection

---

🟩 Phase 4: Advanced & Agentic (Days 21–27)

Focus: Managed RAG services, Knowledge Graphs, and complex orchestration with LangGraph.

Day	Folder	Type	Topic	Key Tech
21	21_rag_as_a_service_ragie/	Streamlit	RAG-as-a-Service	Ragie + Anthropic
22	22_rag_agent_cohere_qdrant/	Streamlit	Multilingual RAG	Cohere + Qdrant
23	23_rag_database_routing/	Streamlit	Dynamic Collection Routing	Multi-tenant Qdrant
24	24_knowledge_graph_rag_citations/	Streamlit	Knowledge Graph Citations	Neo4j + Ollama
25	25_contextualai_rag_agent/	Streamlit	Managed Datastores	ContextualAI
26	26_ai_blog_search_langgraph/	Streamlit	Agentic LangGraph flow	Gemini + Qdrant
27	27_corrective_rag_langchain/	Streamlit	CRAG with Web Search	Tavily + LangChain

---

🟥 Phase 5: Expert & Multimodal (Days 28–30)

Focus: Autonomous research agents, multimodal (Vision) RAG, and production diagnostics.

Day	Folder	Type	Topic	Key Tech
28	28_autonomous_rag_agent/	Streamlit	Autonomous Researcher	PDF + Web Autonomy
29	29_rag_failure_diagnostics_clinic/	CLI	RAG Fatigue Diagnostics	P01–P12 Failure Modes
30	30_vision_rag_multimodal/	Streamlit	Multimodal Vision RAG	Cohere Vision + Gemini

Environment

See .env.example for common environment variables used across tutorials.

Contributing

See CONTRIBUTING.md.

License

MIT — see LICENSE.