Learn
Working notes on the machinery underneath — how these systems actually compute, not how they are usually described. Interactive where interaction earns its keep.
Interactive
State estimation
The extended Kalman filter, without the hand-waving
Where the Jacobians actually come from, why the covariance update is written the way it is, and what breaks when the linearisation is bad. Derived rather than quoted — then fused live from GPS, a wheel encoder and an IMU.
Runnable simulations
Editable filter code
Every step derived
Interactive
Neural networks
A transformer small enough to read
An 18,304-parameter decoder-only transformer trained to memorise the multiplication table — small enough that every matrix, every attention head and every gradient can be inspected while it trains. Step it one batch at a time and watch the loss move.
Live, steppable training
Per-layer inspection
3D activation flow