NUTRITRACK — AI-POWERED NUTRITION & FITNESS TRACKING THE PROBLEM Tracking nutrition manually is tedious, inaccurate, and requires constant guesswork. Users struggle to precisely estimate portion sizes and macro breakdowns, while manually entering meals and syncing workout data across different apps often leads to frustration and abandoned fitness goals. NutriTrack solves this by using AI vision and text processing to automatically analyze meals, alongside a machine learning model that predicts and tailors health insights.

WHAT IT DOES

Analyzes food images and natural language descriptions to instantly extract ingredients, macros, and calories.
Generates a personalized health score using a custom machine learning model trained on diet and biometric data.
Automatically syncs activity and workout data from Strava to provide a holistic view of daily energy expenditure.
Consolidates all health data into a unified, high-performance dashboard with smooth animations.

HOW IT WORKS

Frontend. Next.js and React 19 deliver a fast, server-side rendered interface enhanced with GSAP and Lenis for smooth scrolling.
Inference. The Python Flask backend routes user images and text to Google Gemini (or other LLMs) for intelligent nutritional extraction.
Prediction. A local Scikit-learn model evaluates user diet and activity to generate custom health scores and insights.
Orchestration. Supabase handles secure authentication, while MongoDB Atlas persists user profiles, meal logs, and workout history.

SYSTEM ARCHITECTURE The core engineering approach focuses on a seamless, self-contained deployment model that runs the entire full-stack application reliably. Enforced with:

A single Docker container housing both the Next.js frontend and Flask backend.
Nginx acting as an internal reverse proxy to route traffic seamlessly between the client and APIs.
Secure, environment-based configuration for connecting to all external services (Supabase, MongoDB, Strava, Gemini).

VALIDATION & DEPLOYMENT Configured for production-ready cloud deployment with minimal friction:

Single-service deployment blueprint via render.yaml for one-click setup on Render.
Fallback heuristic mechanisms to ensure the app continues functioning even if external AI providers are unavailable.
Stateless backend API design allowing for horizontal scaling if user load increases.

TECH STACK Next.js 16, React 19, GSAP, Python, Flask, Scikit-learn, MongoDB Atlas, Supabase, Google Gemini API, Docker, Nginx

STATUS AND LINKS Under active development as a full-stack web application.

Source: https://github.com/Mzuhaibkhan/Nutritrac.git
Live Link: https://nutritrack-64xy.onrender.com/