Frigo Hero

Staring at your empty fridge? Snap photos, AI creates zero-waste recipes tailored to your diet, allergies, fitness goals & cooking time. Smart shopping lists save money!

Comment

Have you ever stood in front of your open fridge wondering “What’s for dinner?”, only to find that the answer lies in a single photo that transforms ingredient chaos into the meal of your dreams? 📸🍽️✨

💡 Inspiration

Every day, millions of people worldwide face a recurring Hidden Kitchen Crisis characterized by a frustrating mix of food waste and nutritional guesswork. This problem is underscored by the staggering reality that nearly 1.3 billion tons of food are wasted globally each year, while rising costs of living make household budget management more difficult than ever.

Frigo Hero was born from my desire to bridge this gap by treating the refrigerator not just as a cold storage box, but as a real-time nutritional logistics engine that starts with the user’s actual reality — exactly what is sitting in their fridge and pantry.

I was inspired to transform the mundane chore of “checking the fridge” into a personalized culinary adventure that removes the cognitive load of calculating macros and planning meals. By integrating advanced biometric data such as Age, Height, and Weight with specific fitness goals and critical allergy restrictions, I created an ecosystem that makes healthy, zero-waste eating feel like an inviting Ghibli-inspired experience 🌱🎥 — including a Smart Shopping System and Real-Time Pricing that completes the technical loop.

🧠 Main Functionalities: Precision Nutrition & Smart Logistics 1️⃣ Biometric Integration & Fitness Intelligence 🏋️‍♂️

Biological Analysis: The AI establishes a nutritional foundation by analyzing your specific Age, Height (e.g., 180 cm), and Weight (e.g., 75 kg).

Custom Objectives: Whether your goal is Weight Loss or Muscle Gain, the system automatically adjusts caloric density and portions to match your target.

Zero Manual Math: The AI handles all complex nutritional equations, ensuring every meal serves your body’s specific needs no guesswork required 🧮✨

2️⃣ AI “Fridge-to-Plate” Analysis 🤳🍳

Smart Scan: A single photo of your fridge or pantry allows the AI to instantly recognize available ingredients with high precision.

Constraint Filtering: The system cross-references ingredients with your Allergies (e.g., Lactose, Peanuts) and Dietary Regimes (e.g., Halal, Vegan) to ensure complete safety.

Macro Calculation: Every generated recipe includes a detailed breakdown of Calories, Proteins, and Carbs calculated per serving.

Dynamic Scaling: The AI scales ingredient quantities based on the number of people to ensure Zero-Waste cooking ♻️

3️⃣ Smart Shopping System & Real-Time Pricing 🛒💰

Gap Identification: Any missing essential ingredients are automatically added to a smart grocery list.

Store Suggestions: Based on your geolocation, the app suggests three local stores (e.g., Carrefour, Lidl) where you can find these items.

Real-Time Pricing: The app displays estimated prices in your local currency (e.g., EUR), helping you choose the most economical and healthy options.

🎨 UX/UI Design & Prototyping

Before writing a single line of code, I mapped out the user journey in Figma. I focused on a “Zero-Friction” interface, ensuring that the transition from scanning a fridge to viewing a recipe takes fewer than three taps.

The UI was designed with a calming, Ghibli-inspired aesthetic, using Tailwind CSS to implement a responsive, mobile-first design that adapts perfectly to any screen size 📱✨

🚀 How I Built It

I engineered a robust full-stack architecture designed for performance and security:

Frontend: Built with Angular 18, utilizing Signals for reactive, high-performance state management. I used Ionic and Capacitor to wrap the application into a native mobile experience.

Mobile Deployment: Prepared using Xcode for iOS and Android Studio for Android builds.

Backend Proxy: A Spring Boot (Java 17) server acts as a secure bridge. This ensures a “Zero-Secret Frontend”, hiding all API keys and prompt logic behind secure middleware 🔐

AI Engine: I integrated Gemini 3 to analyze multimodal fridge photos. For visual engagement, I utilized Imagen 4.0 to generate recipe illustrations in a soothing Studio Ghibli style 🎨

Database & Auth: Firebase handles real-time synchronization via Firestore and secure user authentication (Google & Facebook).

🧪 Quality Assurance & Testing

To ensure platform reliability, I implemented a dual-layer testing suite:

Frontend Testing: Karma + Jasmine were used to test Angular Signals and services, verifying that user preferences like allergies and diets are correctly processed.

Backend Testing: Mockito + JUnit 5 were used to mock service layers, ensuring the API Proxy correctly handles Gemini responses and user data without exposing sensitive information.

⚠️ Challenges I Ran Into

The most significant technical hurdle I faced was the Context Complexity problem dynamically scaling recipe quantities based on biometric data while ensuring the AI respected every persistent allergy.

I also navigated the difficulty of forcing a Large Language Model to return strictly structured JSON to prevent parsing errors.

Additionally, implementing a seamless multilingual experience was a major challenge. I ensured the AI could understand prompts and generate accurate culinary instructions and localized pricing across Arabic, French, and English, while maintaining consistent tone and technical precision 🌍

🏆 Accomplishments I’m Proud Of

I’m incredibly proud of achieving a fully secured, production-ready AI architecture. By moving all sensitive calls to a Spring Boot proxy, I ensured that no API keys or system prompts are ever exposed in client code delivering a true Zero-Secret security model 🔒

I also successfully merged nutritional science with a Ghibli-inspired UX, proving that a technically complex AI tool can still feel warm, inviting, and intuitive.

Watching the system identify cluttered ingredients and instantly calculate personalized macros for specific fitness goals was a real “Aha!” moment ✨

📚 What I Learned

This hackathon was a masterclass in Multimodal Prompt Engineering. I learned how to structure complex system instructions to handle diverse ingredients and dietary laws consistently.

I also mastered Angular Signals for state management, proving that modern reactive patterns significantly improve the fluidity of data-heavy AI applications.

🔮 What’s Next for Frigo Hero

My journey doesn’t end with the hackathon 🚀

💳 Subscription Model: I plan to launch a premium tier offering unlimited Ghibli-style image generations and advanced fitness tracking.

🔗 URL-to-List: A feature in development that allows users to paste a recipe video or link, instantly converting it into a filtered grocery list based on their current fridge inventory.

📲 App Store Release: I’m finalizing builds for a global launch on the Apple App Store and Google Play Store.

Built With

  • ai
  • android-studio
  • angular-18
  • angular-signals
  • capacitor
  • ci/cd-pipelines
  • cloud-based
  • computer-vision
  • css
  • facebook-authentication
  • figma
  • firebase
  • firebase-authentication
  • firestore
  • gemini-3-api
  • geolocation-apis
  • git
  • github
  • google-authentication
  • gradle
  • html
  • imagen-4.0-api
  • ionic
  • jasmine
  • java
  • javascript
  • json-schema-enforcement
  • junit-5
  • karma
  • large-language-models-(llms)
  • maven
  • mobile-first-design
  • mockito
  • multimodal-ai
  • price-comparison-apis
  • prompt-engineering
  • responsive-design
  • rest-apis
  • spring-boot
  • tailwind-css
  • typescript
  • xcode
Share this project:

Updates