Qariin

Inspiration

Local communities have massive untapped resources during disasters—trucks, generators, medical training—but lack coordination. We wanted to turn these scattered neighborhood assets into an organized response network for the critical hours before formal help arrives.

What it does

Qariin is a smart disaster response platform. When an emergency triggers, it dynamically matches households needing help with nearby volunteers, resources, and safe community hubs based on urgency, proximity, language compatibility, and available capacity.

How we built it

We prioritized a fast, reliable MVP over shiny tech. We used React (Vite), Mapbox and Featherless API for a QWEN model to handle real-time state changes and built a custom matching algorithm. We strictly used AI for utility: real-time chat translation and request summarization. We built a React frontend where all state is managed client-side with mock data to simulate a real disaster scenario reliably during the demo. The map is powered by Mapbox GL JS, showing live neighbor dots, community hubs, and a dynamic travel radius circle. AI is used in two real ways: Featherless (running Qwen2.5) generates ultra-concise 3-bullet action checklists for helpers, and translates in-app chat messages in real time for cross-language communication.

Challenges we ran into

Balancing ambition with hackathon time constraints. We had to ruthlessly prioritize features, mock complex routing to avoid broken APIs, and design an ethical "vulnerability" tagging system without storing sensitive medical data.

Accomplishments that we're proud of

Building a matching engine that actually feels intelligent, implementing real-time chat translation to break down language barriers, and delivering a flawless, end-to-end demo of a flood scenario.

What we learned

In crisis tech, reliability beats complexity. Designing for degraded connectivity and simple UI flows is far more valuable than adding complex GIS engines or bloated features.

What's next for Qariin

Building offline mesh networking for when cell towers go down, adding SMS fallback for users without data, and implementing AI-driven task splitting for complex, multi-step rescue needs.