AI-driven training platform for Hyrox athletes with personalized workout scaling.
Hyrox athletes train across eight demanding stations. Generic programs ignore individual capacity. Most athletes either undertrain or burn out trying to follow plans built for someone else.
The challenge: design an AI system that understands workout video, extracts form and intensity, then scales difficulty based on real performance—not arbitrary percentages.
Video analysis models that extract form, intensity, and rep quality. Personalized scaling algorithms based on recovery and performance trends.
Detailed technical schematics and documentation for Hyrox Trainer are proprietary and available upon request for deep-dive discussions.
Technical constraints forced creative solutions. We optimized for Fitness AI from day one, which meant rethinking architecture at every layer. Shipped incrementally, validated with real users, and scaled what worked.
Delivered on scope, timeline, and technical requirements
AI products need to be grounded in real constraints. Hallucinations kill trust. Accuracy at scale beats feature bloat.
I'm drawn to hard problems at the intersection of emerging technology and human behaviour especially in spaces that are ripe for disruption powered through innovation.