FitMaster AI

Final Year Project, NUST

Role / Services

AI & Full Stack Development

Credits

Team FitMaster AI

Location & year

Islamabad, Pakistan ©

2024 – 2025

FitMaster AI progress dashboard: daily calories, weight over time, weekly exercise and exercise form accuracy

The project

FitMaster AI is an AI fitness companion that makes expert workout and nutrition guidance affordable: it writes your workout and diet plans, recognises food from a photo to count your calories, and tracks your progress.

I led it as our final year project at NUST, in a team of four with Raahim Shah, Haroon Jawad and Awais Nadeem. It is made for beginners who cannot afford a personal trainer, and as a safe alternative for women who prefer not to train with a male coach.

The front end is Next.js and TypeScript with Tailwind CSS, shadcn/ui and Chart.js, on Firebase for sign-in and data. An OpenAI chatbot gives fitness advice and writes the plans, a YOLO model behind a FastAPI service recognises the food in a photo, and real-time posture detection with MediaPipe and TensorFlow corrects your form as you train.

What I built

  • Workout and diet plans written by an OpenAI chatbot
  • Food recognition from a photo with a YOLO model, for calorie tracking
  • Real-time posture detection with MediaPipe and TensorFlow
  • Progress dashboard with Chart.js
  • Next.js, TypeScript, Tailwind CSS and shadcn/ui on Firebase
  • FastAPI back end for the AI predictions
FitMaster AI food tracking: a photo of a meal recognised as fried rice, ice cream and samosas, with its calories, protein, carbs and fats
FitMaster AI weekly workout plan, with sets and reps for each day