Software for
fast cars

Aero Atlas

How Formula 1's rules shaped the cars, from the 1961 1.5-litre formula to the 2026 active-aero cars, in one interactive 3D model.

• three.js
• JavaScript
• Procedural 3D geometry
• Data research

• Builds a regulation car for every season from that year's FIA figures (width, wing spans and heights, overhangs, wheelbase, tyres, weight), with a cited source on every number and both values shown where sources disagree.
• All geometry is generated in code: lofted chassis and sidepod sections, NACA-style aerofoils, lathed tyres, and era-specific details such as skirts, grooved tyres, the halo and 2026 movable wings.
• "Legal box" view of the space the rules allow, an illustrative airflow view, era comparison as a ghost overlay, and every season's grid from the open F1DB dataset.

Pitwall

An F1 telemetry lab: compare any drivers' laps from real timing data, replay them in 3D, and run a live strategy pit wall.

• Python
• FastAPI
• FastF1
• OpenF1
• Pyodide
• Plotly
• three.js

• Lap comparison from official timing data: speed, throttle, brake and gear traces, lap-time delta, track dominance, braking zones and corner-by-corner analysis. The fastest lap is always the reference.
• Pit wall strategy engine: fuel-corrected tyre-degradation fits, undercut and overcut calls, rejoin position, pit window, safety car and VSC time loss, and an ERS estimate.
• 3D lap replay with chase, onboard and TV cameras; race replays from past data, and support for a live OpenF1 feed within its rate limits.

The online version runs the same Python analysis in your browser (Pyodide) on 8 recorded sessions; run it locally for any session and live timing.

Hybrid Energy-Management Simulator

How should an engine and an electric motor share the load? A simulator that tests energy-management strategies against the best possible one.

• Python
• NumPy
• Dynamic programming
• ECMS

• Quasi-static model of a parallel turbo-hybrid: drag, rolling resistance, Willans-line engine, MGU-K, battery equivalent circuit and an MGU-H.
• Compares rule-based control, ECMS and a dynamic-programming benchmark priced fairly on stored energy. ECMS captures 92–100% of the achievable hybrid saving.
• Checked the class report's claims: 280 km/h and 500 km range are two different operating points, and 500 km holds only up to about 176 km/h steady.

Petrol Turbo-Hybrid Powertrain

A class project to design a vehicle powertrain, taken from concept to CAD model and scale build.

• CAD
• Vehicle dynamics
• Technical report

• Ruled out an early self-charging wind-turbine concept on conservation of energy: pulling energy from the car's own airflow costs more drag than it recovers.
• Sized power at speed from the drag power equation P = ½ρC_dAv³, then iterated through combustion, battery-electric and hybrid layouts.
• Settled on a petrol turbo-hybrid as the best balance of range, performance and feasibility, then built the simulator above to test the report's numbers.