What it does This repo is a massive directory of Jupyter Notebook tutorials organized by difficulty: beginner OCR and chat UIs, intermediate agentic workflows and MCP integrations, and advanced fine-tuning plus production deployments. Each project is a standalone implementation you can run, not a Medium article with pip install commands buried in paragraph four.

The interesting bit The breadth is the point. Rather than one perfect demo, it samples the entire AI engineering landscape — from local Llama OCR to multi-agent deep researchers, from sub-15ms RAG retrieval to building reasoning models from scratch. It’s a map of what “production-ready” currently means in a field that redefines itself weekly.

Key highlights

• 93+ projects sorted into Beginner (22), Intermediate (48), and Advanced (23) tiers
• Heavy emphasis on local/self-hosted options: Ollama, Llama variants, DeepSeek, Qwen
• Strong MCP (Model Context Protocol) coverage — 10+ projects on the emerging standard for tool-augmented agents
• Model comparison section with head-to-head evaluations (Llama 4 vs DeepSeek-R1, Sonnet4 vs Qwen3-Coder, etc.)
• Includes a full “AI Engineering Roadmap” for structured learning

Caveats

• “Production-ready” is the repo’s claim, not independently verified; some projects are clearly tutorial-grade
• No visible CI, testing infrastructure, or dependency lockfiles mentioned — expect breakage as models update
• Newsletter signup is aggressively promoted throughout the README

Verdict Grab this if you’re a developer who learns by breaking working code and wants to survey the full agent/RAG/MCP landscape without building 93 repos yourself. Skip it if you need a single, maintained framework with stable APIs.