Open-source AI engineering curriculum for building intelligent systems.
From ML Fundamentals → LLMs & RAG → AI Agents → Production Deployment. Hands-on lessons with PyTorch, LangChain, LlamaIndex, DSPy, FastAPI & more.
Curriculum • Quick Start • Student Guide • Exam Prep • LinkedIn • Support Project
This repository is free, open-source educational material (MIT). Use it to teach yourself, run a study group or a course; fork it, translate it, send improvements. See CONTRIBUTING.md and the Code of Conduct. If it helps you, please cite it (CITATION.cff) and star the repo.
This is an independent learning project, shared openly as educational material. Use responsibly and follow your institution's academic integrity policies. See Academic Integrity Policy for details.
A comprehensive AI engineering curriculum from ML fundamentals to production AI agents and intelligent systems.
- AI Agents with LangChain, LangGraph, and LlamaIndex
- RAG Systems with vector databases (Pinecone, Weaviate, Chroma)
- Production APIs with FastAPI and modern deployment
- LLM Applications with prompt engineering and DSPy
- Real Projects from notebooks to deployed systems
Basic ML → Deep Learning → LLMs & RAG → AI Agents → Production Systems → Advanced Research
| Structure | Basic 18 · Intermediate 26 · Advanced 15 · Agents 18 · Applied 35 · Expert 26 · Electives 39 = 177 lessons (≈ 500+ h) |
| Every Lesson | real scenario → objectives → concept → build from scratch → use the real tool → break it → exercises (Practice/Stretch/Challenge) → checkpoint |
| Hands-On | runs offline, free, no GPU; agent lessons also run live against Claude with one environment variable |
| Honest | the table below is generated from the files on disk, so it can never promise lessons that don't exist |
| Tier | Lessons | Hours | Ready | Retained | Planned | Folder |
|---|---|---|---|---|---|---|
| Basic: Learn from data | 18 | 48 | 0 | 18 | 0 | Basic/ |
| Intermediate: Vision, sequences, language | 26 | 78 | 0 | 26 | 0 | Intermediate/ |
| Advanced: LLM engineering | 15 | 48 | 15 | 0 | 0 | Advanced/ |
| Agents: AI agent systems | 18 | 56 | 18 | 0 | 0 | Agents/ |
| Applied: Production frameworks | 35 | 140 | 10 | 5 | 20 | Applied/ |
| Expert: Research frontiers | 26 | 92 | 0 | 0 | 26 | Expert/ |
| Electives: Specialized topics | 39 | 117 | 0 | 39 | 0 | Electives/ |
| Total | 177 | 579 | 43 | 88 | 46 |
▸ Foundations ▸ AI Agents ▸ Production
├─ PyTorch ├─ LangChain ├─ FastAPI
├─ TensorFlow ├─ LangGraph ├─ Docker
├─ scikit-learn ├─ LlamaIndex ├─ Kubernetes
├─ NumPy/Pandas ├─ DSPy ├─ CI/CD
└─ Transformers └─ CrewAI └─ Monitoring
▸ Vector DBs ▸ LLMs ▸ Frameworks
├─ Pinecone ├─ OpenAI ├─ Haystack
├─ Weaviate ├─ Anthropic Claude ├─ Semantic Kernel
├─ Qdrant ├─ Open Source LLMs └─ Guardrails AI
└─ Chroma └─ Fine-tuning
git clone https://github.com/nexageapps/AI.git && cd AI
python -m venv .venv && source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install -r requirements.txt
jupyter labOptional, for LIVE agent runs: export ANTHROPIC_API_KEY=sk-ant-... (or put it in a git-ignored .env). Without a key, every agent notebook replays a recorded trace so you can still run and break everything for free. Never commit keys.
Lessons are named <ID>-<kebab-title>.ipynb, e.g. AG03-tools-done-right.ipynb: the ID is a tier letter plus two digits (B Basic, I Intermediate, A Advanced, AG Agents, AP Applied, E Expert), the title is lowercase-hyphenated with no spaces. The last lesson in each tier is the capstone. Full rules in CURRICULUM.md §6; checked by python tools/check_names.py.
| You are… | Do this | Time |
|---|---|---|
| New to ML | Basic → Intermediate → Advanced → Agents | ~220 h |
| Know ML, want to build agents (fast track) | A01 → A04, then AG01 → AG18, then AP01–AP10 | ~110 h |
| Know LLM APIs already | AG02 → AG18 → Applied | ~90 h |
| Want research | Basic (skim) → Intermediate → Advanced A05–A07 → Expert | ~150 h |
Two 3-hour lesson sessions + one 2-hour review/project block. One lesson = one session; the capstone of each tier gets a full week.
| Tier | Folder | Lessons |
|---|---|---|
| Basic · Learn from data | Basic/ |
B01–B18 |
| Intermediate · Vision, sequences, language | Intermediate/ |
I01–I26 |
| Advanced · LLM engineering | Advanced/ |
A01–A15 |
| Agents · AI agent systems | Agents/ |
AG01–AG18 |
| Applied · Production frameworks | Applied/ |
AP01–AP35 |
| Expert · Research frontiers | Expert/ |
E01–E26 |
| Electives · Specialized topics | Electives/ |
Optional |
AI/
├── Basic/ Intermediate/ Advanced/ # 59 lessons: ML fundamentals to LLMs
├── Agents/ Applied/ # 53 lessons: AI agents and production
├── Expert/ # 26 lessons: Research frontiers
├── Electives/ # 39 lessons: Specialized topics
│
├── agentkit/ # Shared library for agent lessons
├── tools/ # Build, test, and curriculum tools
├── application/ # Interactive demos and visualizations
├── documentation/ # Guides, curriculum, and policies
└── README.md LICENSE requirements.txt
Interactive demonstrations of AI concepts through hands-on web applications. All demos are playable in your browser.
Interactive AI demos — playable in your browser
| Demo | Concept | Notebook | Link |
|---|---|---|---|
| Wumpus World | Symbolic logic, knowledge representation | B01 - Symbolic Logic | Play |
| Demo | Concept | Notebook | Link |
|---|---|---|---|
| Mountain Explorer | Gradient descent, optimization | B02 - Linear Regression | Play |
| Blindfold Hill | Learning rate, convergence | B02 - Linear Regression | Play |
| Neural Network Trainer | Forward/backward propagation | B05 - Neural Networks | Play |
| CNN Explorer | Convolution, pooling, feature maps | I01 - CNNs | Play |
| Demo | Concept | Notebook | Link |
|---|---|---|---|
| Data Preprocessing Studio | Missing values, scaling, encoding | B06 - Data Preprocessing | Play |
| Model Evaluation Dashboard | Metrics, confusion matrix, ROC/AUC | B07 - Model Evaluation | TBD |
| Demo | Concept | Notebook | Link |
|---|---|---|---|
| Knowledge Graph Playground | RDF triples, knowledge graphs, RAG | A03 - RAG | Play |
| RNN Explorer | RNN, LSTM, GRU, sequence modeling | I03 - RNNs | Play |
| Transformer Explorer | Self-attention, multi-head attention | I04 - Transformers | Play |
| BPE Explorer | Tokenization, byte pair encoding | I05 - BPE | Play |
View All Demos • Explore application/ for source code and deployment guides.
A lesson is ready only if it follows the template and python tools/nbtest.py "<path>" passes offline. See tools/LESSON_AUTHORING.md. After adding or changing lessons run python tools/gen_readmes.py to refresh the status tables. See also CONTRIBUTING and SECURITY.
- Learning concepts and understanding implementations
- Preparing for lectures and exams
- Using as inspiration for original projects
- Understanding different approaches
- Copying code for assignments without understanding
- Submitting repository code as your own work
- Using during closed-book assessments
- Violating your institution's policies
Full Academic Integrity Policy
Open-source AI education built for learners worldwide.
LinkedIn • Discord • Buy me a book
MIT License: see LICENSE.