Meal-E

Let me look at what I know about Meal-E from our conversations to fill this in well.Here's a filled-in version based on your project:

Inspiration

As college students, we constantly struggle with food waste and meal planning. Ingredients sit in the fridge until they expire, we forget what we have, and we end up spending money eating out instead of cooking what's already available. We wanted to build something that bridges the gap between what's in your fridge and what's on your plate — especially for students on a budget trying to eat healthy and hit their nutritional goals.

What it does

Meal-E is a smart meal planning system that uses a fridge-mounted camera to automatically detect ingredients via computer vision, stores them in a pantry database with nutritional info and expiration dates, and generates personalized weekly meal plans using AI. It prioritizes ingredients that are expiring soon, respects dietary preferences and allergies, and optimizes for user-defined macro targets (calories, protein, etc.). The system includes a FastAPI backend, a PostgreSQL database for pantry management, and a Swift iOS frontend.

How we built it

We used computer vision (via a fridge camera script) to identify ingredients entering and leaving the fridge. The backend is built with FastAPI and PostgreSQL (via pg8000) to sync pantry state, track nutritional data, and manage expiration dates. Meal plan generation is powered by Gemini's API, prompted with the user's pantry inventory, dietary preferences, and macro targets. The iOS frontend is built in Swift, giving users a clean interface to view their pantry and weekly meal plans.

Challenges we ran into

Getting the pantry sync right was tricky — we hit issues with column naming mismatches in our database schema, date parsing errors from passing numeric values into date fields, and managing the correct parameter ordering in our SQL inserts. Integrating the AI meal plan generation required careful prompt engineering to ensure the model only used ingredients actually in the pantry and accurately calculated macros. Setting up the Git workflow across the team also had its bumps with divergent branches and repo restructuring mid-hackathon.

Accomplishments that we're proud of

We built a fully functional end-to-end pipeline in 24 hours — from a camera detecting ingredients, to a database tracking them, to an AI generating meal plans from them. The system actually respects dietary constraints, prioritizes expiring food, and produces meal plans with real recipes and accurate macro breakdowns.

What we learned

We learned a lot about prompt engineering for structured JSON output from LLMs, working with PostgreSQL in Python, handling edge cases in computer vision pipelines, and coordinating a full-stack project across a 5-person team under time pressure. We also deepened our understanding of nutritional data modeling and how to build reliable data flows between hardware, backend, and frontend.

What's next for Meal-E

High-quality barcode scanners and internal cameras for improved accuracy Production-grade vision models for reliable detection Edge + cloud integration for fast processing and scalability Automated tracking with minimal user effort Adaptive freshness modeling based on real-time temperature data