The idea: tools sorted by concept depth — how much you need to understand to really
use each one — from Level 0 (use ready-made AI, no theory) to Level 4 (research- and
production-scale systems). It's not about how hard a tool is to click; ChatGPT is easy to use (Level 0)
even though the ideas behind it are advanced.
Every tool has two links: 📖 Read — an approachable article/tutorial to understand it —
and 🐙 Code — its repo or homepage. Levels are judgment calls; you might move a tool ±1 level, and
that's fine.
Awesome AI/ML Tools — 211 tools by concept depth (Level 0 to 4)
An awesome list of AI and machine-learning tools for LLMs & Generative AI — searchable and leveled by concept depth (Level 0 to 4). Every tool links to an approachable article to read and to its code or homepage. Enable JavaScript for search, autocomplete and level/category filters.
Level 0 — Consume ready-made GenAI
Use finished AI products. No code, no theory.
Chat assistants
Coding assistants
Media generation
Research assistants
Level 1 — Run & build with no/low code
Run models locally via a GUI, or assemble apps by drag-and-drop. First ideas: prompts, tokens, RAG.
Local model runtimes
No-code / low-code builders
Access & studios
Level 2 — Code with high-level libraries
Write Python against SDKs and batteries-included frameworks; embeddings, basic orchestration, demos.
Model provider SDKs / APIs
Orchestration frameworks
Models & embeddings
App / demo UIs
Structured output
Protocols & starter agents
Level 3 — Build, fine-tune & ship
Engineer RAG/agent systems, fine-tune models, use vector DBs, evaluate, and operate (LLMOps, serving).
Advanced agent frameworks
RAG frameworks & ingestion
Vector databases
Fine-tuning
Evaluation
Guardrails & security
Observability & LLMOps
Serving & gateways
Media generation (power-user)
Embeddings & reranking
Level 4 — Research & scale
Distributed training, high-throughput inference engines, GPU kernels, quantization internals, RLHF at scale.
Distributed / large-scale training
High-throughput inference / serving
GPU kernels, quantization & efficiency
Research evaluation / benchmarks