Stage 015 lessons
Mathematical foundations
Vectors, calculus, optimization, probability, and the learning process—explained as tools, not obstacles.
Full AI Course
A personalized, plain-English course from the mathematics of machine learning to deep learning, AI agents, knowledge graphs, ontologies, and emerging AI systems.
One field. One connected learning path.
Full AI Course
Learn the whole story of AI in simple English: the mathematics underneath it, the code that makes it real, and the modern systems that turn models into agents.
- 47 core chapters
- 3 frontier capstones
- 100s of visuals and demos
AI
understand → build → deploy
∑ Math
</> PyTorch
⬡ Graphs
◎ Agents
01Plain EnglishBuild the mental picture first
02Visual modelSee the idea move and connect
03MathematicsRead every symbol in words
04PyTorchTurn the formula into working code
Your course, not a generic syllabus
Tell the course how you want to learn
Choose three things. Your answers stay in this browser and shape the recommended order, pace, and next lesson.
Your recommended route
Foundation-first explorer
A course that remembers
Your progress
The complete map
Eight connected stages, 50 lessons
Follow the complete sequence or let your profile move the most useful stages to the front. Nothing is locked.
Recommended nowComplete
Stage 028 lessons
Classical machine learning
Prepare data, train reliable models, measure them honestly, and understand structured uncertainty.
Stage 035 lessons
Deep learning
Neurons, backpropagation, CNNs, sequences, attention, transformers, and generative models in PyTorch.
Stage 0410 lessons
Applied AI
Vision, language, audio, time series, LLMs, multimodal AI, reinforcement learning, and real-world systems.
Stage 053 lessons
Production & infrastructure
Deploy, monitor, scale, optimize, and govern real AI systems instead of stopping at a notebook.
Stage 067 lessons
Reasoning, knowledge & safety
Search, logic, planning, ontologies, causal thinking, interpretability, ethics, fairness, and safety.
Stage 076 lessons
Graphs, robotics & frontier systems
Graph neural networks, knowledge graphs, robotics, learning theory, retrieval, and building LLMs from scratch.
Stage 086 lessons
Post-training & modern AI capstones
Fine-tune, align, serve, and connect models into safe agentic and knowledge-grounded systems.
One idea, four representations
The formula is never left floating
Every important equation is translated into ordinary language, a picture, and executable code. Here is the core of one neuron.
Score, shift, then squash
Multiply each input by its importance, add them, add a bias, then squeeze the result into a useful range.
\(y = \sigma(Wx + b)\)
W holds learned weights, x is the input, b shifts the decision, and σ adds non-linearity.
import torch
layer = torch.nn.Linear(3, 1)
x = torch.tensor([[0.8, 0.2, 0.5]])
y = torch.sigmoid(layer(x))
print(y)Stack this idea
One neuron makes a small decision. Millions or billions of these operations, connected in layers, create modern deep-learning models.
Beyond the standard ML syllabus
Modern topics, connected to their foundations
01
Tool-using agents
Planning, memory, tools, handoffs, guardrails, MCP, A2A, and evaluation.
02Graphs & ontologies
Triples, schemas, GNNs, knowledge graphs, GraphRAG, and grounded reasoning.
03Reasoning & world models
Test-time compute, model-based agents, embodied AI, and continual learning.
04Frontier watch
A dated, evidence-linked view of what is becoming practical—and what remains uncertain.
Start small. Keep the whole map.