KindBites

KindBites turns food waste into hope. We connect restaurants with surplus meals to people in need—using AI to ensure safety, trust, and speed. Fighting hunger with every donation.

Comment

Inspiration

Every year in the U.S.,millions of pounds of perfectly edible food go to waste—84.3% of unused restaurant food is discarded, while just 1.4% is donated. At the same time, food insecurity affects millions of families. Our team was inspired to bridge this gap.

With KindBites, we set out to build a smart, accessible solution that empowers restaurants to donate food with confidence, connects those in need to local supplies, and improves food sustainability.

What it does

KindBites is a web platform that connects restaurants with surplus food to individuals and food shelters in need. Here's how it works:

  1. Restaurants upload/take pictures of their available produce.
  2. A computer vision (CV) model using HuggingFace API scans the food image to assess freshness and safety, preventing spoiled or unsafe food from entering the system.
  3. Verified listings are then posted on a marketplace, where users can browse, claim, and arrange pickups of available food.

This ensures trust, accountability, and simplicity for both donors and recipients.

How we built it

KindBites consists of 3 core components:

  • Marketplace: A central hub where restaurants post verified food items and users can browse and claim donations.
  • Educational Articles: A learning space that helps users recognize signs of food spoilage and understand safe food handling.
  • RAG Chatbot: A retrieval-augmented generation chatbot that answers real-time questions about food safety and sustainability tips using a custom MongoDB knowledge base.

We combined HuggingFace and Streamlit for CV modeling, and a RAG-based NLP framework to make the experience interactive and intelligent.

Challenges we ran into

  • Equity and Accessibility: Designing for users with lower digital literacy or language barriers required us to think critically about how to make KindBites inclusive and usable for all.
  • Deploying RAG: Since Ollama is designed for localhost we had trouble finding a way to deploy for everyone to interact with our chatbot. This is a future implementation we would like to do.

Accomplishments that we're proud of

  • Built a dual-access platform that serves both donors (restaurants) and recipients (individuals and shelters), making food redistribution more accessible.
  • Integrated a real-time CV model to verify food safety—building trust and ensuring quality for all users.
  • Designed an educational ecosystem through articles and an AI-powered chatbot to raise awareness about food sustainability and empower everyday action.

What we learned

  • Oziel learned to integrate the Hugging Face API for spoilage prediction, display pickup locations using the Google Maps Embed API, and handle image data in MongoDB with base64 encoding.
  • Richa Through this project, I learned how to style sections using custom HTML and CSS within Streamlit, organize content into columns, and dynamically update the page based on user selections.
  • Asmitha enhanced her skills in building a RAG chatbot with Python leveraging Langchain for document preparation and MongoDB for semantic search.

What's next for KindBites

To increase accessibility and scale our impact, we would like to:

  • Add GPS-based navigation, helping users find the nearest pickup locations quickly and efficiently.
  • Gamify sustainability actions, rewarding users who regularly donate or collect food.
  • Introduce freshness and location filters to enhance the food browsing experience.
  • Implement a cart system, allowing users to add food items for organized pickup and checkout.
  • Deploy the RAG chatbot using a Groq API so that it is accessible for everyone and not just through locally.

Built With

Share this project:

Updates