The Problem
42 million Americans face food insecurity. That's 1 in 8 people.
But here's what shocked me: most people struggling with malnutrition don't have access to quick, reliable tools to:
Assess if they're nutritionally at-risk
Find nearby food banks or community resources
Get meal plans that work with what they can actually afford
I wanted to fix that. So I built NutriAI.
What is NutriAI?
NutriAI is an AI-powered platform that helps people in underserved communities assess malnutrition risk and connect with food resources nearby.
Key features:
AI Risk Assessment - Machine learning model that evaluates malnutrition risk based on demographics, symptoms, and dietary patterns
Food Resource Map - Interactive map showing food banks, community kitchens, and affordable markets nearby
Smart Meal Plans - AI generates personalized meal suggestions based on available ingredients and budget
Community Exchange - Share surplus food with neighbors, request items you need
Why I Built It (And Why You Should Too)
When I started this project, I wasn't thinking "how do I build something impressive?" I was thinking "what problem actually matters?"
Food insecurity is one of those problems that seems too big to solve alone. But then I realized: I can make a tool that helps thousands of people assess their own risk and find resources. That's impact.
The hackathon deadline (Fuel the Future 2026) forced me to actually finish something instead of endlessly tinkering.
The Tech Stack
I wanted to build something that worked great, deployed easily, and didn't require a backend (easier to launch, no server costs).
Here's what I used:
Frontend: React 18 (loaded via CDN + Babel Standalone)
3D Effects: Three.js (animated torus knot + particle effects)
Styling: Custom CSS with glassmorphism
Fonts: DM Sans, Space Grotesk (Google Fonts)
Deployment: GitHub Pages (free!)
Why these choices?
React via CDN - No build process needed. I could write JSX and have it work immediately. Perfect for rapid prototyping.
Building the Features
- AI Risk Assessment This was the core feature. I built a machine learning evaluation that takes user inputs:
Age, location, income level
Symptoms (fatigue, weakness, frequent illness, etc.)
Dietary patterns (how often they eat vegetables, protein, etc.)
The data comes from a combination of Google Places API and community databases. Users can filter by resource type and distance.
- Smart Meal Plans This was fun. The AI takes:
Available ingredients
Nutritional needs (based on risk assessment)
Budget constraints
Dietary preferences
And generates meal plans with recipes. It prioritizes nutrient density and affordability.
- Community Exchange This is the social layer. Users can:
Post surplus food they have
Request items they need
Connect with neighbors
Build community resilience
It's simple but powerful—helps bridge the gap when people have extra and neighbors need it.
Challenges I Hit (And How I Solved Them)
Challenge 1: No Backend = No User Data Storage
Problem: I wanted users to save their meal plans and preferences, but I had no backend to store data.
Solution: Browser localStorage. It's not fancy, but it works for a prototype. Users' data stays on their device, privacy is protected, and it still feels like the app "remembers" them.
What I'd Do Differently (Next Time)
Start with backend earlier - localStorage works, but a real database would enable features like user accounts, community features, etc.
User testing sooner - I built based on assumptions. Talking to 3-5 people from the target community earlier would have shaped priorities differently.
Simpler first version - I could have launched with just the risk assessment + resource map. The meal planner and community exchange can come later. Done > Perfect.
Documentation from day one - I scrambled to write the README at the end. Would've been easier to document as I built.
Key Takeaways
For builders:
Ship something imperfect - NutriAI isn't perfect, but it exists and people are using it
Choose boring tech - React, Three.js, GitHub Pages. Nothing bleeding-edge. Boring = reliable = shipped
Build for a real problem - This project stands out because it solves something real, not because of fancy tech
For problem-solving:
Big problems need small solutions first - Food insecurity is huge. I can't fix it alone. But I can give people a tool to help themselves. Start small.
Accessibility matters - Mobile-first, fast-loading, works with slow connections. The people who need this most have the worst internet.
Try It Out
Live site: https://kalrakrish777-ctrl.github.io/nutriai/
GitHub: https://github.com/kalrakrish777-ctrl/nutriai
I'd love feedback! What features would be most useful? Any bugs? Any ideas for improvement?
Top comments (0)