AI Sport Tech Consultant
Andrea
Zignoli
I help sport tech startups turn physiology and performance data into production-ready AI solutions that drive subscriptions, enhance UX, and deliver new features.
I bridge the gap between sports science research and deployed code — whiteboarding with exercise physiologists in the morning, shipping containerised APIs by evening.
Background
Who I am
Research & Education
M.Eng. Mechatronics + PhD Sports Science — I enrolled to study robotics and ended up studying the most wonderfully made machine: the human body.
Post-doctoral research at the University of Trento (sport performance modelling), Yokohama National University (rehabilitation robotics), and the University of Verona (exercise physiology).
I maintain active research collaborations and publish in peer-reviewed journals. Associate Editor at Sports Engineering (Springer) and peer reviewer for several top sports science journals.
Clients & Collaborations
AI-driven training platform
Modelling, feature design, backend development
Services
What I do
Training Prescription & Assessment
Adaptive algorithms that personalise training load based on individual athlete physiology, historical performance data, and real-time feedback signals.
Cardiopulmonary Exercise Testing
ML models for automated threshold detection and CPET interpretation — from ventilatory thresholds to VO₂ kinetics — deployed at scale.
Continuous Glucose Monitoring
Insights extraction from CGM data for metabolic optimisation in endurance athletes — identifying fuelling patterns, glycaemic responses, and actionable recommendations.
Pacing & Race Strategy
Simulation models for optimal pacing, incorporating aerodynamics, terrain, fatigue dynamics, and competitive constraints to maximise performance.
Approach
How I work
I combine first-principles modelling with machine learning. Depending on the problem, I write differential equations by hand, train neural networks, or blend both approaches.
I prototype fast, validate against real data, and iterate until the model delivers business value. Full-stack: from research and model development to Docker containers, Flask/FastAPI services, and production deployment.
Recent work: educational courses, performance models deployed as APIs, web apps connected to CMS via LLM pipelines.
Tech stack
Side work
Personal projects
Research
Recent publications
Speaking
Talks & podcasts
Deep dive on how AI agents are being integrated into training platforms, how HRV readiness assessments are evolving, and what the next generation of training methodologies looks like.
Why traditional models like W′ fall short, and how Workout Reserve offers a more flexible mechanical way to understand athlete capability. The future of AI coaching: RAG-AI, LLMs, and agents in training platforms.
Why modeling must come before AI can be useful in coaching. Models like Critical Power, Banister TRIMP, and HRV response curves provide the structure AI needs to make smart, adaptable decisions.
HRV and its role in optimising training readiness and recovery. As co-architect of Athletica's AI-assisted HRV monitoring, I explain how AI enhances HRV analysis while maintaining the importance of human expertise.
Development and application of Workout Reserve — an AI-driven tool that helps athletes measure their current effort against historical performances — with real-world applications in elite cycling and Ironman.
Contact
Work together?
Is there a project idea you want to discuss? Whether it's a research collaboration, a custom model, or a tool you'd like to build — I'm always happy to explore new ideas. Send me a message and let's talk.
andrea.zignoli@unitn.it