NutriLink

NutriLink AI uses face scanning, AI chat, voice access, and smart food matching to fight malnutrition—delivering the right nutrients to underserved communities, not just meals.

Comment

NutriLink AI — Delivering the Right Nutrients, Not Just Food

Inspiration

Over 700 million people worldwide suffer from malnutrition—not just hunger, but a lack of essential nutrients like protein, iron, and vitamins.

What shocked us most is that even when food is available, it often doesn’t match what people’s bodies actually need. Food distribution systems focus on quantity, not nutritional quality.

We wanted to change that. So we built NutriLink AI—a platform that uses AI to ensure people receive the right nutrients, not just calories.

What It Does

NutriLink AI is a mobile-first platform that combines AI, accessibility, and food redistribution to tackle malnutrition in a smarter way.

AI Nutrition Scanner

  • Uses computer vision to analyze facial indicators
  • Estimates potential malnutrition risk (non-diagnostic)
  • Suggests nutrients and affordable foods

Nutrition AI Assistant

  • Chat-based assistant powered by AI
  • Provides simple, low-cost, locally relevant food recommendations
  • Answers nutrition-related questions

Voice Assistant

  • Speech-to-text + AI + text-to-speech
  • Enables hands-free interaction
  • Designed for low-literacy users

Smart Food Redistribution

  • Connects donors (restaurants, stores) with receivers (NGOs)
  • Matches food based on nutritional needs, not just availability
  • Reduces food waste while improving health outcomes

Impact Dashboard

  • Tracks meals delivered
  • Tracks nutrients distributed
  • Measures real-world impact

How We Built It

We combined multiple technologies into one unified platform:

  • Frontend: React + React Native (Expo)
  • Backend: Firebase / Supabase

  • AI & APIs:

    • OpenAI (chat + reasoning)
    • Whisper (speech-to-text)
    • Text-to-Speech APIs

  • Computer Vision: TensorFlow.js / MediaPipe

  • Data Handling: Cloud database + local caching for low connectivity

We also used heuristic-based models for malnutrition screening due to limited datasets, ensuring the system remains lightweight and fast.

Challenges We Ran Into

  • Limited training data for accurate malnutrition detection
  • Balancing AI complexity vs. hackathon time constraints
  • Designing for low-resource environments (low bandwidth, low literacy)
  • Ensuring the system is scalable and realistic, not just theoretical

What We Learned

  • Accessibility is just as important as innovation
  • AI doesn’t need to be perfect to be impactful
  • Real-world constraints lead to better, more thoughtful design
  • Combining multiple simple systems can create powerful solutions

What’s Next

  • Improve AI accuracy using real health datasets
  • Add offline-first syncing for rural communities
  • Expand multilingual support
  • Partner with NGOs for real-world pilot programs
  • Enhance food matching using more advanced optimization models

Our Mission

To make nutrition accessible, intelligent, and equitable for everyone—regardless of location or resources.

Built With

  • api
  • apis
  • css
  • firebase
  • mediapipe
  • native
  • openai
  • react
  • supabase
  • tailwind
  • tensorflow.js
  • text-to-speech
  • whisper
Share this project:

Updates