🤝 Top Collaborators
🤓 Latest Submissions
CamSort AI
Security teams managing large facilities face a real problem. When you have hundreds or even thousands of cameras, it is impossible for a small team to monitor all of them at once. Most camera management software just shows you a wall of feeds and leaves the prioritization entirely up to the human operator. CamSort solves this by introducing an autonomous AI agent that continuously analyzes camera snapshots and ranks them based on what the operator actually cares about at that moment. The operator types a natural language instruction such as "sort by how hazardous the scene appears" or "find cameras with people present" and CamSort takes it from there. The system sends each camera snapshot independently to Google Gemini, which returns a score, a classification, a reason, and a recommended action for each feed. The backend then sorts all cameras by score and updates the dashboard automatically every few seconds without any manual input. CamSort also includes an operator priority layer that flags any camera crossing a critical threshold, making sure urgent situations are never buried under lower priority feeds. The live dashboard shows ranked camera cards with scores, reasons, and operator actions so the team always knows exactly where to look first. The project is built on Vultr, uses Google Gemini for multimodal image analysis, and is deployed with a live demo accessible to judges.
19 May 2026
QuantTrade Lite
QuantTrader Lite is a fully autonomous AI-powered crypto trading agent designed to solve three core problems in modern trading: information overload, lack of trust in AI decisions, and slow human reaction time. Built for the Lablab.ai Hackathon 2026 (Kraken CLI Track), the system continuously fetches live market data, analyzes it using Groq AI (Llama 3), and executes paper trades via Kraken CLI—all without human intervention. 🧠 How It Works The system follows a 5-step autonomous pipeline: Market Data Ingestion Fetches real-time BTC price and 24h change from CoinGecko API. AI Decision Engine Groq-powered Llama 3 analyzes market trends and generates a BUY / SELL / HOLD signal along with a clear, human-readable explanation. Trade Execution The decision is executed using Kraken CLI in sandbox mode (paper trading). Logging & Transparency Every action is recorded in a structured trade_log.json file for auditability. Live Dashboard A Streamlit interface displays signals, trade history, and charts with auto-refresh every 60 seconds. 💡 What Makes It Different Explainable AI Every decision includes a clear reason—no black-box trading. Fully Autonomous Runs continuously with zero human input. Hackathon-Compliant Direct integration with Kraken CLI ensures full alignment with challenge requirements. Simple but Powerful Built entirely in Python with a lightweight, production-ready architecture. 🛠 Tech Stack Groq API (Llama 3) → AI decision-making Kraken CLI → Trade execution (sandbox) CoinGecko API → Live market data Streamlit → Real-time dashboard Python 3.11+ → Core system 🎯 Impact QuantTrader Lite transforms crypto trading from manual, overwhelming, and opaque into a system that is: ⚡ Fast 🔍 Transparent 🤖 Autonomous It not only trades—but also teaches users why each decision is made, bridging the gap between AI and human trust.
12 Apr 2026
