AI & ML — MATHS FOR AI/ML

Maths for AI/ML

Most topics here have three levels — Introduction, Intermediate, and Advanced — so pick one and go as deep as you want. A few go straight to Intermediate for readers who already have the basics down.

Vectors & Dot Products
The arrow-in-space math underneath embeddings and attention — what a vector is, what the dot product measures, and why both matter for AI.
Introduction Intermediate Advanced
Matrix Multiplication
Why a transformer layer’s forward pass is fundamentally repeated matrix multiplications, worked by hand, and why GPUs are built specifically to run it fast.
Introduction Intermediate Advanced
Probability & Softmax
How raw output scores become a real probability distribution over the next token — and how temperature, top-k, and top-p pick from it, worked in real numbers.
Introduction Intermediate Advanced
Gradient Descent & Loss Functions
How training a model actually works — measure how wrong it is, nudge the weights downhill, repeat millions of times.
Introduction Intermediate Advanced
Backpropagation & the Chain Rule
How the gradient that gradient descent needs actually gets computed, layer by layer, using the chain rule.
Introduction Intermediate Advanced
Statistics for Model Evaluation
The mean, variance, and confidence interval math underneath every eval score — and why sample size changes how much a "Model A beats Model B" claim is worth.
Introduction Intermediate Advanced