beautiful architecturing

🧠 Project Inspiration

Food waste is one of the most pressing sustainability challenges of our time. As university students juggling busy schedules and tight budgets, we saw an opportunity to create something meaningful—an app that not only reduces waste but also empowers users to eat sustainably, healthily, and affordably.

We realized that many restaurants throw away perfectly good food at the end of the day, while students like us are searching for inexpensive, convenient meals. At the same time, dietary restrictions—whether due to allergies, religion, or lifestyle—are often overlooked in existing food waste solutions.

This sparked Replate—a generative AI-powered platform that connects users with surplus food from restaurants while personalizing recommendations to their preferences and restrictions. Our goal was to create a system that understands each user's unique needs and recommends safe, sustainable meals in real time.

⚙️ Technology Stack

🧑‍💻 Languages

TypeScript
JavaScript
Python

🧰 Frameworks and Libraries

Node.js – Backend infrastructure
React & Next.js – Interactive frontend
Leaflet.js – Dynamic mapping of radius of preference
Express.js – Node.js backend API
Flask – Python backend for AI coordination
Twilio – Communication and SMS notifications
Socket.io – Real-time streaming and delivery tracking
Gemini API – Natural language interpretation
Cohere API – Contextual RAG and preference processing
Too Good To Go API – Restaurant and surplus food data sourcing

☁️ Platforms / Cloud Services

Google Cloud Platform (GCP) – Hosting backend services and deploying AI agents
Gemini API – Understanding user intent and requests
Cohere API – Semantic analysis and preference interpretation
Twilio – SMS messaging and notification system

🛠️ Tools and Software

Postman – API testing and endpoint verification
Liveshare – Real-time collaborative coding
Git & GitHub – Version control and team collaboration
Node.js Runtime – Local development and server support

🚀 Product Summary

Replate is an AI-powered food sustainability and delivery app that connects users to surplus food from local restaurants—curated to match individual preferences, dietary restrictions, and sustainability values. It combines the sustainable practices of Too Good To Go with the convenience of Uber Eats. You get the food that you wish you could eat, that restaurants wish they could sell.

👨‍🍳 Key Features

Personalized Meal Discovery: Users input food preferences, allergies, and religious restrictions. A multi-agent AI system—powered by Gemini and Cohere—interprets their intent and filters the available food inventory accordingly.
Generative AI Filtering: Instead of relying on rigid JSON tagging, our system semantically infers appropriate meals using generative AI.
Real-Time Recommendations: Leveraging Socket.io, users receive live updates on their order through the client interface.
Sustainable Sourcing: Replate connects with local restaurants offering surplus meals.
Green Future Vision: We envision a future with robot-powered food collection and delivery, pushing toward zero-emission logistics and zero-food waste.

🧠 AI Innovation

Our system introduces a multi-agent generative AI architecture. Gemini and Cohere engage in a reasoning loop, exchanging context until they agree on the user's intent and the best meal recommendation. This consensus-based system emulates human reasoning and handles the high-context complexity of food preferences and sustainability in a scalable, automated way.