SaveMePls is a Decentralized Swarm Intelligence Drone Rescue System designed for disaster response. During the critical first 72 hours of a crisis, rescue efforts are crippled by communication blackouts, human fatigue, and a lack of manual drone pilots. Our solution is a fully autonomous, 100% offline drone swarm. Powered by Local LLMs at the edge and Agentic AI, the system automatically partitions search zones, orchestrates multi-drone pathfinding, and locates victims without requiring internet connectivity or human intervention, drastically reducing search times when traditional infrastructure collapses.

AntiFault is an advanced Digital Twin and Predictive Maintenance Platform that eliminates costly industrial downtime by transforming raw machinery telemetry into immersive, actionable 3D visualizations. Addressing the inefficiencies of reactive maintenance, the system utilizes AI to calculate the Remaining Useful Life (RUL) of components, predicting hardware failures before they occur. The platform’s standout feature is Neo, an integrated Voice AI Assistant that allows operators to monitor equipment health and automate complex scheduling tasks through natural language, such as saying, "Hey Neo, schedule maintenance for the hydraulic press." By bridging the gap between complex data analysis and intuitive floor operations, AntiFault provides a high-tech "bird's-eye view" that enhances both safety and operational efficiency in industrial environments.

Project Beacon is a robust Tauri-based Ground Control Station designed for search-and-rescue in comms-denied environments, bridging the gap between flexible AI reasoning and deterministic drone control. The system utilizes a multi-agent orchestration layer that interprets natural language commands via a Next.js and FastAPI architecture, offering a hybrid "BYO API" online mode or a fully offline local LLM mode. By leveraging FastMCP to expose drone tools and gRPC for executing commands within Docker-simulated environments, it enables fleet-level coordination, self-correcting safety behaviors like battery management, and real-time 3D telemetry visualization via React Three Fiber. Ultimately, Project Beacon provides operators with a reliable, high-tech command center that maintains situational awareness and auditable mission logs even when traditional infrastructure fails.

Digital Fraud Shield is a multi-layered safety ecosystem designed to protect elderly and rural ASEAN communities from sophisticated digital scams through a "Defense-in-Depth" strategy. By shifting away from traditional "Hard Blocks" that often fail against social engineering, the system utilizes Edge AI for private, real-time call analysis and an AI Risk Engine that introduces "Adaptive Friction"—such as mandatory native-language quizzes—to break a scammer’s psychological spell. The platform further secures high-risk transactions through the Guardian Protocol, requiring family member approval for large transfers, and provides an AI-Powered Recovery tool to generate forensic reports for law enforcement. Ultimately, Digital Fraud Shield replaces digital isolation with proactive oversight, combining high-tech precision with human intervention to secure the region’s most vulnerable users.

SIREN is an offline AI rescue drone orchestration system that solves the "day zero" connectivity gap in disaster response. While traditional AI systems fail when infrastructure collapses, SIREN runs a local Qwen 3.5 9B model with zero internet dependency to manage autonomous drone swarms in real time. Using a LangGraph-based Command Agent and the Model Context Protocol (MCP), it autonomously partitions search sectors, self-heals the swarm if units go offline, and provides commanders with an explainable Chain-of-Thought audit trail. By combining on-device LLM reasoning with thermal heatmap detection, SIREN delivers mission-critical tactical intelligence to agencies like NADMA Malaysia exactly when the grid goes dark.