Optimal Primers

Part 1: Linear Regression

A hands-on introduction to the building blocks of neural networks

Part 2: The Perceptron

Building out the classic classification model, and sharpening our maths

Part 3: The Multi-Layer Perceptron

Modelling non-linear data from scratch, and unlocking universal approximation

Part 4: Automating the Chain Rule

Unlocking composability by building primitives

Part 5: The Convolutional Neural Network

Sliding windows applied to neural nets

GPT Shakespeare

A character-level transformer trained on Shakespeare, running live in the browser via ONNX

Beard or Badger

A scratch-built CNN image classifier built in vanilla JS — no frameworks, no GPU, just convolutions