01 // projects

Projects

Things I've built. Each one has a short write-up here and the full code on GitHub.

Career stripes: one coloured stripe for each of Max Verstappen's Grands Prix

33 → 1 → 3

Max Verstappen's Formula 1 career as a data story, told through his three car numbers. Charts, not predictions.

He lost the 2025 title by 2 points, two years after winning 19 of 22 races.

data visualization python plotly
write-up github
The F1 Race Predictor app comparing predicted and actual finishing positions

F1 Race Predictor

Predicts the finishing order of a Grand Prix from qualifying and form. Machine learning, not a story.

Got the exact podium in 42% of 2025 races, against 33% for "finish where you start".

machine learning scikit-learn
write-up github live app
Replay of the trained agent and a fixed-time signal running the same traffic

Adaptive Traffic Signal

A Q-learning agent that runs a traffic light in the SUMO simulator, deciding every 5 seconds whether to switch.

63.7% less queuing than a fixed-time signal, on traffic it never trained on.

reinforcement learning sumo
write-up github live demo
Draw a Digit app reading a hand-drawn 8

Neural Network from Scratch

A 784 → 128 → 10 network in plain NumPy that reads handwritten digits. Every gradient written by hand.

Read my own badly drawn 8 with 99.0% confidence.

numpy deep learning
write-up github