MealMatch – AI for Sustainable Meal Planning

Hackathon

Sustainability

Next.js

Generative AI

OCR

AWS Bedrock

Amazon Textract

Tailwind CSS

START Lausanne Hackathon • March 2025

Product & Impact Strategist

Project Overview

Participated in START Lausanne Hackathon 2025 in a team of three, developing a platform that turns scanned grocery receipts into multi-day meal plans. The app promotes sustainable consumption and reduces food waste through ingredient-based meal generation using Claude 3.5 Sonnet (AWS Bedrock) and Amazon Textract.

Challenges

•Building a working prototype within 48 hours using a modern full-stack tech stack
•Handling OCR for diverse receipt formats and mapping to usable ingredient structures
•Integrating generative AI for meaningful, waste-minimizing meal suggestions
•Designing a responsive, intuitive UI for rapid input and plan customization
•Ensuring data consistency and usability within a serverless PostgreSQL backend

Key Achievements

•Deployed a functional web app on Vercel with full receipt-to-meal workflow
•Integrated OCR and generative AI for real-time, tailored suggestions
•Used Neon and Next.js to build a lightweight, scalable serverless platform
•Demonstrated strong impact potential and clear usability for real-world audiences

Technologies Used

Next.js 13

TypeScript

Tailwind CSS

AWS Bedrock

Amazon Textract

Neon (PostgreSQL)

Vercel

View Project