ASSETHUNTER // LAVAPUNK
"The old systems are cooling, solidifying into obsidian. It's time to inject new heat."
AssetHunter is a "Beast Mode" Campus Recovery Protocol built for the LAVAPUNK hackathon. It replaces chaotic WhatsApp groups and fading paper notices with a high-speed, AI-powered aggregation engine.
🔥 The Mission
Campus lost-and-found is broken. It's slow, analog, and fragmented. AssetHunter centralizes this data into a tactical feed using:
Computer Vision: To instantly tag and categorize items.
Geospatial Grid: To visualize loss density across campus sectors.
Speed: Optimistic UI updates and a dark, "Obsidian" aesthetic designed for power users.
⚡ Key Features
👁️ AI Sentinel Scanner: Snap a photo, and our local AI (Ollama/Moondream) instantly analyzes it. It extracts the title, tags, color, and condition—and is smart enough to know a pen isn't "Electronics."
🗺️ Tactical Radar Map: A custom OpenStreetMap implementation (Web) and Native Map (Mobile) that visualizes found items as signal blips on a dark-mode grid.
🚀 Live Feed: An optimized, flat-list stream of assets with "Lost" (Orange) and "Found" (Cyan) status indicators.
🔐 Universal Access: Runs seamlessly on Android, iOS, and Web (PWA).
🛠️ Tech Stack
Framework: Expo (SDK 54) + React Native
Web Support: React Native Web
Language: TypeScript
AI Engine: Ollama (Running moondream locally)
Maps: react-native-maps (Mobile) / OpenStreetMap Embeds (Web)
Styling: Custom "LavaPunk" Theme System
⚙️ Setup & Installation
- Prerequisites
Node.js (LTS)
Ollama installed on your computer.
Expo Go app installed on your phone.
- Clone & Install
git clone https://github.com/aaryanpawar16/Asset-Hunter.git
cd AssetHunter
npm install
- 🧠 The "Brain" Setup (Ollama)
This app uses a Local LLM for privacy and speed. You must configure Ollama to accept connections from your phone.
Pull the Vision Model:
ollama pull moondream
Configure Network (Windows PowerShell):
Step A: Right-click the Ollama icon in your taskbar and select Quit. (Critical!)
Step B: Run this command to open the server to your local network:
$env:OLLAMA_HOST="0.0.0.0"; $env:OLLAMA_ORIGINS="*"; ollama serve
Note: Ensure ollama.exe is allowed through Windows Firewall (Private & Public).
Update App Config:
Open src/services/ai/ollama.ts.
Find const LOCAL_IP and replace it with your computer's IP address (run ipconfig to find it).
🚀 Running the App
Start the development server:
npx expo start --clear
📱 Mobile: Scan the QR code with the Expo Go app.
💻 Web: Press w in the terminal to open the Tactical Dashboard in your browser.
🏆 Hackathon Notes (Judges)
Why Web? We prioritized a "Universal Protocol". Students shouldn't need to install an APK to find their keys. The Web version features a custom CSS-filtered map to maintain the cyberpunk aesthetic without Google API keys.
Why Local AI? Privacy. We process images on the edge (or local server) rather than sending student data to a generic cloud API.
The "Error" Handling: If the AI fails (or network drops), we have a robust fallback that allows manual entry, ensuring the system never freezes.
"Sometimes we need the beast to come out like RAHH."