PitchPulse
AI-powered sports science assistant for player readiness, injury risk, vitals, movement analysis, and tactical XI planning.
Project information
- Category: AI Sports Science / Hackathon
- Dates: Feb 2026
- Tech: Flutter, Swift, FastAPI, Python, Firebase, Supabase, Presage SDK, Gemini 2.5 Flash, Actian VectorDB, Cloudflare Tunnels
- GitHub: PitchPulseAI
- Devpost: PitchPulse
Project Details
Overview
PitchPulse is an AI-driven high-performance management system for football managers. It combines match workload, player check-ins, rPPG vitals, movement screening, and GenAI recommendations into a single readiness and tactical planning workflow.
Problem
Professional teams lose major value when player availability drops because of preventable workload spikes and soft-tissue injuries. Managers often have physical match data, but not a unified view of emotional readiness, biometric signals, movement quality, and tactical risk.
Engineering
- Built a Flutter mobile app with Provider state management and a Swift bridge for the Presage SDK to extract heart rate, HRV, stress, and emotional state from a selfie check-in.
- Engineered a Python/FastAPI backend that calculates Acute:Chronic Workload Ratio and updates risk bands from match statistics, vitals, and movement analysis.
- Used Gemini Vision and Gemini 2.5 Flash to produce structured JSON action plans, suggested XI decisions, and mechanical movement corrections from player context.
- Used Firebase authentication, Supabase storage, Actian VectorDB retrieval, and Cloudflare Tunnels to support the mobile-to-backend workflow.
Impact
Shipped a full-stack, multimodal sports science platform in 36 hours with readiness dashboards, injury-risk explanations, recovery recommendations, and game-day squad planning.