EdgeWatch: AI Fleet Telemetry

EdgeWatch is an AI-powered fleet monitoring system designed to detect IoT device problems before they become failures. The project simulates six ESP32 edge devices using Wokwi, with each device periodically reporting telemetry such as temperature, voltage, vibration, signal strength, and uptime. Two devices intentionally develop abnormal behavior: one experiences a gradual voltage drop while another overheats. A Python-based ingestion layer collects the telemetry, maintains a rolling history of recent readings for every device, and stores the data for analysis. Instead of sending individual requests for every device, EdgeWatch batches the fleet's recent telemetry into a single AI analysis request. An LLM analyzes the combined data, identifies anomalies, classifies device health, explains potential problems, and predicts possible failures. The resulting analysis is saved as structured snapshots that can power a live monitoring dashboard. The goal is to demonstrate a practical AI infrastructure pipeline connecting simulated edge hardware, telemetry ingestion, rolling-window processing, centralized AI reasoning, and actionable predictive maintenance. EdgeWatch shows how intelligent monitoring can scale beyond individual devices to entire fleets.