Skip to content
Back to Projects

Food ResQ: Smart Recipes from Fridge Scraps

Hack The Valley 8Oct 2023 - Oct 2023

Led the product design and UI/UX for Food ResQ, an AI-powered fridge management app developed during Hack The Valley 8 to combat household food waste. Designed complete user experience for tracking fridge inventory via MongoDB database and generating flexible recipe recommendations using ChatGPT. Created intuitive interface for quickly adding ingredients (even unmeasured or unknown foods), built using React.js and Bootstrap frontend with Flask backend. Managed API integration with OpenAI for consistent recipe generation and wrote detailed prompts to retrieve formatted cooking instructions. Secured 6th place overall and won Best Use of MongoDB Atlas, demonstrating ability to deliver polished demo under 36-hour hackathon constraints.

Skills & Technologies

React.jsFlaskMongoDBOpenAI APIPrompt EngineeringUI/UX DesignFigmaBootstrapProduct ManagementHackathon

AI-Powered Recipe Recommendations to Reduce Food Waste

Food ResQ uses a MongoDB database to track ingredients in your fridge and generates flexible, delicious recipes using ChatGPT to help you finish leftover food before it spoils. Built in 36 hours during Hack The Valley 8, the platform leverages AI to transform neglected ingredients (like half a carrot, half an onion, or a quarter pound of ground pork) into creative meal ideas. The app excels at handling unknown foods and ingredients you're too lazy to measure precisely, using LLM flexibility to generate practical recipes with step-by-step instructions. Users can input their fridge contents, and the AI recommends recipes that prioritize ingredients closer to expiry, ensuring nothing goes to waste. This project demonstrates how combining databases, AI APIs, and thoughtful UX design can solve real household sustainability challenges.

From Personal Frustration to Solution

I was cooking at home one day and kept noticing we had half a carrot, half an onion, and like a quarter pound of ground pork lying around all the time. More often than not it was from me cooking a fun dish that my mother would have to somehow clean up over the week. So I wanted to create an app that would help me use those ingredients I had neglected, so that even if both my mother and I forgot about them, we would not contribute to food waste. This personal frustration became the spark for Food ResQ. In Canada, over 58% of food produced is wasted (35.5 million tonnes annually, worth $49 billion), while 1 in 7 Canadians experiences food insecurity. What started as a solution to finish leftover ingredients in my fridge evolved into a platform that tackles food waste at both household and community levels through AI recipe recommendations and food redistribution.

Key Features

Fridge Inventory Database

MongoDB database stores user fridge contents with flexible schema supporting varied ingredient types, quantities, and expiry dates. Users can quickly add ingredients without precise measurements ("half a carrot", "some ground pork") or even unknown foods. The database tracks what's in your fridge over time, providing the foundation for smart recipe recommendations.

Built with flexibility in mind because home cooking is messy. Users don't want to weigh every ingredient or look up exact names. The loose schema accommodates real-world fridge chaos while still enabling intelligent recipe matching.

ChatGPT-Powered Recipe Generation

OpenAI API integration generates creative, practical recipes using whatever ingredients you have on hand. Detailed prompt engineering ensures recipes include step-by-step instructions, cooking times, and substitution suggestions. The AI handles inconsistent ingredient data gracefully, turning "half an onion" and "quarter pound pork" into delicious meal ideas.

LLM flexibility is the killer feature. Traditional recipe databases require exact ingredient matches, but ChatGPT improvises brilliantly with partial ingredients, unknown quantities, and unusual combinations. This makes the app genuinely useful for real kitchens.

Expiry Priority Algorithm

Smart recommendation system prioritizes ingredients closer to expiry, ensuring older food gets used first. When generating recipes, the AI receives expiry context and weights suggestions toward ingredients about to spoil. Future feature: automated notifications reminding users to cook specific ingredients before they go bad.

Prevents the "out of sight, out of mind" problem. Users forget what's buried in their fridge, but the app remembers and actively suggests recipes before ingredients spoil. This proactive approach drives higher engagement and real waste reduction.

Receipt Scanning Vision (Planned)

Planned feature to automatically add ingredients via receipt scanning, eliminating manual data entry. Users snap a photo of their grocery receipt, and computer vision extracts item names and quantities to populate the fridge database. Makes onboarding frictionless and encourages consistent inventory tracking.

Manual ingredient entry is the biggest friction point. Receipt scanning removes this barrier entirely, making the app as easy as "take a photo after shopping." Planned for next iteration to drive adoption and daily usage.

React + Flask Architecture

Clean separation between React.js/Bootstrap frontend and Flask Python backend. Frontend handles user interactions and recipe display, while backend manages MongoDB operations, OpenAI API calls, and prompt engineering. Professional dev-ops practices with Kanban board planning and detailed API documentation prevented refactor headaches during the 36-hour sprint.

Good architecture saved the hackathon. After an initial database schema mistake, the modular design allowed complete refactor without breaking everything. Documented APIs and clear separation of concerns meant the team could parallelize work efficiently.

Tools & Technologies

React.js & Bootstrap

Built responsive frontend with modern component architecture. React handled state management for fridge inventory and recipe display, while Bootstrap provided mobile-friendly styling and UI components. Team had to relearn React during the hackathon, making solid component design critical for staying on schedule.

Flask (Python Backend)

Python Flask server managed API routes, MongoDB database operations, and OpenAI integration. Chose Flask for rapid prototyping speed and Python's excellent library ecosystem for data handling. Backend refactor mid-hackathon tested Flask's flexibility and modular design principles.

MongoDB Atlas

NoSQL database stored user fridge inventories with flexible schema supporting varied ingredient formats. Won "Best Use of MongoDB Atlas" prize for demonstrating how document database flexibility enables real-world messy data (unmeasured quantities, unknown foods, partial ingredients) without rigid schemas.

OpenAI API (ChatGPT)

Core feature powering recipe generation. Engineered detailed prompts to retrieve formatted recipe data including ingredients, steps, cooking times, and substitutions. Learned nuances of writing prompts for consistent LLM outputs and handling API inconsistency challenges during testing.

Figma

Designed complete UI/UX for fridge inventory management, ingredient addition flow, and recipe recommendation display. Created high-fidelity mockups to align team vision before development, preventing feature creep and ensuring focused 36-hour execution.

Kanban Board & Documentation

Professional dev-ops practices with task planning board and detailed API documentation. After initial database schema failure, the team stopped, documented everything, and planned properly. This saved countless hours debugging and enabled efficient parallel development.

Hackathon Success and Technical Learnings

Food ResQ secured 6th place overall at Hack The Valley 8 and won Best Use of MongoDB Atlas for demonstrating how NoSQL flexibility handles real-world messy data. The judges praised our polished demo, professional presentation, and practical approach to household sustainability. Submitting for Best AI Hack and Best Sustainability Hack themes positioned the project at the intersection of emerging technology and environmental impact.

From a technical perspective, Food ResQ taught critical lessons about good API design and planning. We messed up our database schema early on, requiring a complete refactor mid-hackathon. This failure forced the team to drop immediate work, think through solutions together, and document everything properly. The experience reinforced that upfront planning saves massive headaches during implementation. We also learned the nuances of CORS technology when connecting frontend to backend, and discovered how to write detailed prompts for retrieving consistently formatted data from LLMs despite API inconsistency challenges.

The team's professional dev-ops practices made the difference. Our Kanban board saved hours during task planning and implementation. After the initial failure, we documented APIs thoroughly and maintained clear separation between React frontend and Flask backend, allowing parallel development without stepping on each other's toes. The accomplishment we're most proud of: we finished a stunning demo that actually works. Future plans include receipt scanning for automatic ingredient entry, expiry-based notifications, and freemium monetization with premium LLM access for less than a coffee per month. This project demonstrates ability to rapidly prototype AI-powered solutions, recover from technical setbacks, and deliver polished products under extreme time constraints.