Distributed Productivity Telemetry System (IoT/Edge)
A resilient, containerized edge-computing solution for real-time focus analytics and behavioral intervention.
Project Overview
Engineered a distributed microservice architecture to track and optimize productivity by capturing real-time telemetry from client browsers. This Edge-First system processes and stores sensitive user data on a private, containerized Raspberry Pi server and delivers zero-downtime updates via an automated CI/CD pipeline that bypasses local firewalls.
Key Features
- Edge-First Architecture: Keeps user data on a local Raspberry Pi 4 for full privacy and no cloud storage costs.
- Self-Healing Infrastructure: Docker restart policies and systemd daemons ensure automatic recovery after failures.
- Automated CI/CD: GitHub Actions with a Self-Hosted ARM Runner enables push-to-deploy inside private networks.
- Real-Time Intervention: A Manifest V3 Chrome Extension injects DOM overlays based on server analytics.
- Persistent Storage: Data persisted via Docker Volumes (mapped to /data) to protect against container updates.
Technical Implementation
The system is orchestrated with Docker on a Raspberry Pi edge server running FastAPI for telemetry ingestion, SQLite (persisted to volumes) for lightweight storage, and a Vue.js dashboard for visualization. A Chrome Extension collects heartbeats and metadata and sends them to the local backend. The CI/CD pipeline triggers a rolling update on the Pi via a self-hosted runner.
Results & Impact
- 100% Deployment Automation: Updates reduced from ~15 minutes to <60 seconds via Git Push.
- Data Sovereignty: All behavioral telemetry remains on-premises within the LAN.
- High Availability: Recovery under 30 seconds after power cycles using systemd + Docker policies.
- Scalability: Containerized design allows seamless migration to cloud VMs with no code changes.
Project Gallery
Future Enhancements
- Multi-Node Orchestration: Add K3s to manage multiple Pi nodes for redundancy.
- Advanced Visualization: Migrate dashboard to Grafana + Prometheus for time-series insights.
- AI Analysis: Run a local lightweight LLM to produce weekly productivity summaries from logs.