nibble

Front Page of Nibble
Recipe Choices
Weekly Meal Prep Plan
Capture Your Fridge
Nutrients of Meals

Inspiration

You have a midterm tomorrow, 3 assignments due, and you still need to commute home in the rain. The last thing you want to figure out is what to eat, let alone how to get it. For most students, that's when meal plans fall apart: we skip meals, order takeout we can't afford, or eat whatever's left in the fridge. We built nibble help students with their meal prep. You will always know what you'll eat each day, what fits your diet and health goals, how to prepare each dish, and what you can make from what's already in your fridge.

What it does

We built nibble help students with their meal prep. You will always know what you'll eat each day, what fits your diet and health goals, how to prepare each dish, and what you can make from what's already in your fridge.

How we built it

We built the app in React Native so users can enter food preferences, request meals, and upload photos of their fridge. The Python backend sends fridge images to Gemini, which identifies visible ingredients, estimates their quantities, and can generate recipes with cooking instructions. It then sends recipe ingredients to CalorieNinjas to estimate calories and nutrients, which the app displays as approximate per-ingredient or per-serving values. PostgreSQL stores user profiles and preferences so they can be used in meal planning.

Challenges we ran into

Some challenges we faced was when we were making the feature where you can upload a photo of your fridge, and the app will tell you which meals you can cook with the items you have. We initially wanted to use VisionAI, but it could only identify and label what each item was. We ultimately switched to Gemini AI, which could identify what we had in our fridge and how much of it we had.

Accomplishments that we're proud of

We're proud that Nibble actually works: you choose a dish, capture a photo of your fridge, and the app tells you whether you've got what you need. The first time it correctly picked out ingredients from a cluttered, badly lit, very real student fridge was the highlight of our hackathon. Nibble works on both iOS and Android from a single React Native codebase, and we spent real time refining our OpenAI prompting and response parsing so the ingredient check feels reliable instead of like a guess. Most of all, we built something for people like us: uni students trying to meal prep on a budget with limited time and a fridge full of half-used ingredients. We scoped tightly and worked well as a team, and we're walking away with a polished, demo-ready app we'd genuinely use ourselves.

What we learned

While working on this project, we learned how to call APIs, how to convert JSON to readable python, LLMs, databases, prompting, and version control. We also learned how to delegate different parts of the project, and made sure that integrating the front and back end of the project at the end went smoothly.