I mapped ~150 AI/ML tools by "concept depth" — each with a link to read and a link to the code

Most "AI tool" lists are either a wall of 500 links or SEO listicles. I wanted something we all could actually learn from, so I built a leveled map.

🔗 Live: https://maneesh-kumar-thakur.github.io/self-serve-learnings-4-all/
⭐ Code: https://github.com/maneesh-kumar-thakur/self-serve-learnings-4-all

The idea

~150 LLM & Generative AI tools sorted by concept depth, not popularity:

• Level 0 — use ready-made AI, no theory (ChatGPT, Perplexity)
• Level 1 — run/build with no code (Ollama, Flowise)
• Level 2 — code with libraries (LangChain, HF Transformers)
• Level 3 — build, fine-tune & ship (PEFT, vector DBs, LangGraph)
• Level 4 — research & scale (vLLM, DeepSpeed, FlashAttention)

Every tool has two links: 📖 Read (an approachable article to understand
it) and 🐙 Code (the repo or homepage). Nothing AI-generated — the links come
from real docs, tutorials, and courses, and a weekly GitHub Action re-checks all
~300 of them so they don't rot.

Under the hood

One static HTML page (no framework, no tracking), client-side search +
autocomplete + filters, data in a single file, CC0.

What's next

Classical/core ML, Computer Vision, and MLOps are the next topics. What tools am I missing, and where would you move something up or down a level?