FitCoach – LLM-Powered Virtual Personal Trainer with Real-Time Wearable Feedback
Project Description
Create an AI fitness coach that combines LLMs and real-time smartwatch data (e.g., heart rate zones, motion sensors) to adjust workout intensity, prevent overtraining, and provide other related suggestions.
Supervisor
ZHANG Qian
Quota
2
Course type
UROP3100
UROP3200
UROP4100
Applicant's Roles
Understand the usage of LLM and its application in a vertical domain problem space.
Applicant's Learning Objectives
1) Build a system that analyzes live workout data (e.g., maximum/minimum/variation in HR) and provides voice/text guidelines via LLM (e.g., "Slow down—your HR exceeds 85% max!").

2) Generate post-workout summaries with improvement tips (e.g., "Increase cool-down duration based on elevated post-exercise HR").

3) Compare recommendations with certified trainer advice for validation.
Complexity of the project
Moderate