Nightmare Shield
Nightmare Shield is a fully localized, distributed security and monitoring lab designed to resurrect legacy hardware and integrate it with modern systems without relying on cloud APIs. We built a Network Operations Center (NOC) using Netdata to unify a mesh of devices across different architectures and OS distributions (CachyOS, Proxmox, Void Linux, Alpine) via Tailscale. For the Security Operations Center (SOC), we developed Sentinel Shield: a custom agent that pulls telemetry from this diverse cluster. The core of our project leverages local AI inference. Using an AMD Ryzen 7 260 processor, the system runs the Phi-3 model entirely locally—without a dedicated GPU—to analyze infrastructure logs and detect threats in real-time. Nightmare Shield demonstrates how AMD's native architecture can secure hybrid environments at the edge with maximum privacy and zero latency.
Category tags:
"Nightmare Shield is an impressive localized security and monitoring solution that addresses a real need - securing hybrid environments with privacy and zero latency. The fact that it runs Phi-3 locally on a basic AMD Ryzen 7 260 (without a GPU!) for real-time log analysis is remarkable. Using Netdata for NOC and custom Sentinel Shield agent for SOC shows solid engineering. The distributed mesh across different architectures (CachyOS, Proxmox, Void Linux, Alpine) via Tailscale is creative. Privacy-first, zero-cloud approach is a strong differentiator. Application of Technology: 🚀🚀🚀🚀🚀 5 - Runs Phi-3 model locally on basic CPU (Ryzen 7 260, no GPU!) for real-time threat detection. Netdata for unified NOC across mesh devices. Tailscale for distributed networking. Custom Sentinel Shield agent for SOC telemetry. Diverse architecture support (multiple OS distributions). Very creative use of limited hardware. Presentation: 🚀🚀🚀🚀 4 - Clear problem statement (legacy hardware, privacy concerns). Good explanation of the distributed architecture. Links to GitHub and YouTube demo provided. Well-documented implementation. Business Value: 🚀🚀🚀🚀🚀 5 - Addresses critical need for privacy-first security monitoring. No cloud APIs means zero data leakage. Useful for enterprises, individuals, and edge environments. Legacy hardware resurrection saves costs. Originality: 🚀🚀🚀🚀🚀 5 - Very original. Running local AI inference on basic CPU is innovative. Distributed mesh across different architectures is unique. Zero-latency, privacy-first security is a strong differentiator. First-of-its-kind for distributed edge security. "
Sanem Avcil
For Innovators & Creators