Embedded Systems · Spy-o-T Hackathon
Project ALARM — Covert Audio Surveillance System
A distributed embedded keyword-detection security system packaged within a custom-designed digital clock enclosure. The system streams live audio from ESP32-based sensor nodes to a Python backend via WebSockets, where AI-based keyword spotting algorithms trigger automated alert protocols. Designed and fabricated the enclosure to integrate dual ESP32 boards, power distribution, passive cooling, and structured mounting architecture within strict geometric constraints.
This competition strengthened the ability to integrate embedded hardware, real-time communication pipelines, and structured software systems into a cohesive, testable architecture in a fast-paced build environment.
ESP32 (C++, I2S) · SolidWorks · React.js · Python (FastAPI, WebSockets, Telegram API)
Hardware & Enclosure Design
- Designed and fabricated a custom digital clock enclosure in SolidWorks to house dual ESP32 boards with integrated power distribution, passive cooling strategy, and structured mounting architecture within tight geometric constraints.
- Implemented deterministic firmware logic for low-latency I2S audio acquisition, interrupt-aware streaming, and robust WebSocket data throughput under continuous operation.
- Managed power distribution and thermal constraints for a multi-board embedded system operating continuously in a compact, enclosed form factor.
Software Stack & Communication Pipeline
- Built a real-time audio streaming pipeline from ESP32 sensor nodes to a Python FastAPI backend over WebSockets, sustaining low-latency delivery under sustained load.
- Integrated AI-based keyword spotting on the backend to classify incoming audio streams and trigger automated Telegram alert protocols with timestamped event feedback.
- Developed a React.js frontend for system monitoring and remote command execution, providing a live view of detection events and system state during competition operation.
