Recyvo

AiChatBot_LightMode
ProfilePage_LightMode
LeaguePage_LightMode
HomePage_LightMode
ProfilePage_LightMode
ForgotPasswordPage_LightMode
LoginPage_LightMode
MapPage_LightMode
MapPage_LightMode
ProfilePage_LightMode
ScannerPage_LightMode
ScannerPage_LightMode
SettingsPage_LightMode
SettingsPage_LightMode
SettingsPage_LightMode
SignUpPage_LightMode
SignUpPage_LightMode
StartPage_LightMode
AiChatBot_DarkMode
ProfilePage_DarkMode
ProfilePage_DarkMode
ForgotPasswordPage_DarkMode
HomePage_DarkMode
LeaguesPage_DarkMode
LoginPage_DarkMode
MapPage_DarkMode
MapPage_DarkMode
ProfilePage_DarkMode
ScannerPage_DarkMode
ScannerPage_DarkMode
SettingsPage_DarkMode
SettingsPage_DarkMode
SettingsPage_DarkMode
SignUpPage_DarkMode
SignUpPage_DarkMode
StartPage_DarkMode

Inspiration

The core inspiration behind Recyvo came from a universal problem: global recycling rules are fragmented, confusing, and constantly changing. Even with the best intentions, citizens stand in front of public sorting bins completely paralyzed—unsure if a greasy pizza box goes to cardboard or trash, or where to dispose of complex objects like old lightbulbs or alkaline batteries.

Statistics show that over 60% of correctly intended waste ends up contaminating landfills simply due to sorting mistakes. We wanted to build a seamless, AI-driven infrastructure that removes this friction, making waste management accessible, geo-localized, and highly rewarding on a global scale.

How we built it

Recyvo is engineered using a robust, full-stack hybrid architecture optimized for both lightning-fast front-end delivery and intelligent backend orchestration: Frontend: Developed with React, TypeScript, and Next.js, packaged inside a highly scannable, mobile-responsive user interface using Tailwind CSS. Backend: Powered by Python, handling rapid internal API structures, routing logic, and state management. Database Management: Built upon an optimized SQLite database, managed and continually audited via DataGrip during development. Core AI Integration (Gemini API): When a user triggers the scanner, a base64 image data payload is passed through our Python backend directly to the Gemini API along with custom contextual prompt guidelines for immediate visual analysis and waste classification. Interactive Mapping: Built an active tracking module utilizing OpenStreetMap integration to render real-world coordinate layers for localized recycling bin deployment.

Challenges we faced

Handling AI Ambiguity: Computer vision models can naturally struggle with mixed-material items or low-lighting environments, creating a risk of faulty sorting advice. We had to implement a strict prompt-level confidence guardrail: if the model experiences structural uncertainty, it refuses to guess, safely fallback-prompting the user to retry or seek help. Database Orchestration: Merging asynchronous cloud AI callbacks with localized user progress databases while ensuring minimal response latencies required precise database normalization within DataGrip. Scalable Data Crowdsourcing: Setting up a reliable mapping ecosystem that accepts user-submitted locations without vulnerable spam or false reporting required building a pending-queue panel designated for manual human verification.

What we learned

Building Recyvo taught us how to effectively bridge high-level foundational Vision Models (Gemini API) with localized, user-centric web applications. We deeply explored dynamic state-handling in Next.js and data structuring within SQLite. Most importantly, we learned that technology alone doesn't solve environmental crises; rather, it is the deliberate pairing of intelligent AI automation with gamified behavioral hooks (like leagues and rankings) that converts temporary interest into long-term civic habits.

What's next for Recyvo

We plan to significantly expand our API pipeline to allow local commercial enterprises, municipal bodies, and eco-conscious brands to sponsor our competitive Leagues. This framework will allow users to directly trade their accumulated recycling points for real-world municipal transit discounts, utility micro-coupons, and retail vouchers, ultimately closing the loop of a truly circular global economy.