Anticipating system bottlenecks before they crash your workflow.
Aegis Node is a proactive, localized resource management engine and simulated peer-to-peer compute grid. Built for DevGathering 2K26, it shifts the paradigm from reactive monitoring to predictive load balancing using offline, local AI.
Modern development workflows (Local LLMs, multi-container Docker builds, heavy IDEs) have outpaced standard consumer hardware. Aegis Node solves this by:
- Predicting the Crash: A background daemon feeds real-time hardware telemetry (CPU/RAM/Thermals) into a quantized local ML model (phi3:mini via Ollama).
- Privacy-First AI: Inference runs 100% offline. Your proprietary code and system metrics never leave the host machine.
- Autonomous Offloading: When a 90%+ probability of memory exhaustion is predicted, the system automatically routes the pending heavy process to an available "idle node" on a localized P2P community network.
This repository contains the three core microservices of the Aegis Node ecosystem:
- /aegis-cli: The "Watcher." A background Node.js/TypeScript daemon that monitors system health and communicates with the AI layer.
- /aegis-api: The "Router." An Express.js mock controller that simulates the P2P grid, manages node registration, and handles offload requests.
- /aegis-dashboard: The "Command Center." A Next.js (App Router) visual analytics dashboard that renders real-time telemetry and network status.
- Frontend: Next.js, TailwindCSS, Recharts, Framer Motion.
- Backend: Node.js, Express.js, Socket.io (Real-time Full-Duplex).
- CLI Daemon: TypeScript,
systeminformation. - AI/ML: Ollama (
phi3:minitime-series forecasting). - Database: Supabase (PostgreSQL).
- Package Manager:
pnpm.
ollama pull phi3:mini- Start the Backend:
cd aegis-api && pnpm install && pnpm run dev
- Start the Dashboard:
cd aegis-dashboard && pnpm install && pnpm run dev
- Launch the CLI Watcher:
cd aegis-cli && pnpm install && pnpm start
Aegis Node is built on the principle of Responsible AI. By leveraging local inference, we ensure that telemetry data remains strictly on the user's hardware. In a future production environment, all P2P tasks will be executed within isolated, secure Docker sandboxes.
Built by Team Aegis (Dev Sharma & Sambhav Sharma) for DevGathering 2K26.