PaceGuard AI

For long-distance runners, building base endurance and pushing toward milestones like a half-marathon requires immense consistency. However, pushing too hard or ignoring hidden fatigue often leads to the runner's silent enemy: overtraining. Standard fitness apps simply log mileage and pace, but they fail to proactively warn athletes when their training load becomes dangerous. Enter PaceGuard AI, a next-generation intelligent running coach built to predict and prevent injuries before they occur. By analyzing critical sports science metrics—specifically the Acute:Chronic Workload Ratio (ACWR), mileage growth patterns, and Heart Rate Drift—the system accurately detects signs of overtraining and computes an Injury Risk Score (0-100). What sets PaceGuard AI apart is its highly optimized, hybrid AI architecture, purposefully designed for the AMD Hackathon. To maximize computational efficiency, the system employs a "Zero-Token Decision" routing mechanism. Simple data queries and basic interactions are processed locally using AMD hardware, minimizing latency and cost. When a user requires deep, empathetic analysis—such as contextualizing a high risk score or adjusting a weekly training plan—the query is dynamically routed to advanced Cloud LLMs (Fireworks.ai / Gemini). Wrapped in a minimalist, interactive Streamlit dashboard, PaceGuard AI transforms raw running data into clear, actionable, and personalized coaching advice. It doesn't just track your runs; it actively guards your progress, ensuring you can train safely and continuously break your personal records.