Maths4DL was delighted to host this year’s Machine Learning for Differential Equations (ML4DE) Hackathon at the University of Cambridge (15/16 April 2026). Making this edition truly special, we partnered with the National Air Traffic Services (NATS) to craft two compelling challenge problems at the frontier of machine learning and optimal control in flight navigation.

The event drew together more than 20 participants from 15 institutions across four countries – undergraduate and PhD students, postdoctoral researchers and industry professionals alike. Organised into five teams, participants tackled two real-world industrial challenges. In Challenge 1, teams built machine learning-powered “virtual pilots” tasked with steering an aircraft from New York to London while minimising fuel consumption and flight time under uncertain wind conditions. In Challenge 2, teams tackled an open-ended data analysis problem, using real flight tracking data and weather forecasts to investigate why transatlantic flights deviate from their planned routes – and whether smarter routing could help.

Over two days, the hackathon spanned an exciting range of mathematics and machine learning – from the classical elegance of Zermelo’s navigation problem to cutting-edge reinforcement learning. Beyond the technical work, the event buzzed with collaboration, networking, and lively knowledge exchange between academia and industry; all fuelled, naturally, by plenty of pizza.

The hackathon is a vivid demonstration of Maths4DL’s commitment to building practical skills in applied mathematics and machine learning, and to forging meaningful connections between early-career researchers and the industrial partners tackling the challenges of tomorrow.

Huge congratulations to Team “Vibes” and Team “Euclidean Fliers” for taking home the wins, and a heartfelt thank you to everyone who participated. Special thanks to our colleagues at NATS, and in particular Dr Aditya Gaur, for their invaluable partnership and enthusiasm throughout.