Performance Modeling & Interface Design
In collaboration with MIT’s 2.98 Sports Lab, I joined a cross-functional student group of engineers and analysts challenged with how elite cycling teams optimize strategy. We partnered with AusCycling, Australia’s national cycling organization, to develop a real-time optimization model and interface for the 4,000m Team Pursuit race event.
My Role
I led the design integration and the coach experience portion of the project, ensuring the model was not only functional but also intuitive and coach-ready. I was responsible for translating raw performance models, designing not only how the system looked and behaved, but also how it was understood and used by coaches and riders. Incorporating key metrics like power output, W′ depletion (anaerobic capacity), and velocity acceleration to present cyclist performance in a format that was organized and usable in real-time.
Key Design Elements:
Visual Performance Modeling: Created overlays and visuals comparing constant vs. variable power models, highlighting differences in anaerobic strain and acceleration phase tradeoffs.
Interface Architecture: Co-developed two user modes:
On-the-ground coaching view for rapid simulations during track testing.
Pre-race strategy tool for deeper optimization, powered by cloud compute and genetic algorithms.
Modeled race output based on rider W′ values, energy constraints, and track dynamics.
Used genetic algorithm optimizations to test over 60,000+ combinations of team orders, peel points, and pacing strategies.
Achieved total optimization runtimes as low as 4 hours, with interface feedback in under 0.5 seconds.
Sponsor Communications: Created a final handoff to AusCycling stakeholders, producing a sponsor-facing walkthrough video, a narrative UI deck, and interactive documentation.