SensorSentinel

AI-powered agentic testing platform that validates IoT device management systems using intelligent test agents orchestrated through UiPath Automation Cloud.

Comment

Inspiration

My background is in embedded systems and C development. I've always worked close to the hardware, writing firmware, reading sensor datasheets, and debugging low-level communication protocols. When I discovered this hackathon, I saw a unique opportunity to bridge two worlds that rarely meet: the embedded/IoT space and enterprise agentic AI testing.

The inspiration came from a real problem I've observed. IoT systems are notoriously difficult to test. Sensors produce unpredictable data, edge cases are extreme (999°C spikes, voltage dropouts to 0V, negative humidity), and traditional testing approaches simply can't simulate the chaotic nature of real embedded device behavior. I wanted to build something that solves this.

What I Learned

  • How to integrate embedded C code with modern cloud platforms
  • How to design and build agentic test agents using Python and pytest
  • How UiPath Automation Cloud orchestrates workflows and governs agents
  • How to build real-time dashboards with live data pipelines
  • How AI-assisted development with Claude Code accelerates complex builds

How I Built It

SensorSentinel has four core components:

1. C IoT Device Simulator Built in C using WinHTTP and an XOR-shift RNG for optimal performance. The simulator runs in two modes. Normal mode generates realistic sensor readings, while chaos mode deliberately injects all 5 anomaly types: temperature spikes (999°C), temperature underflow (-166°C), humidity overflow (>100%), negative humidity (-1%), and voltage dropout (0V).

2. Python Flask REST API A thread-safe API using a bounded deque store for O(1) operations. Receives sensor data, validates it, tags severity levels (ok, warning, critical), and exposes 5 endpoints for ingestion, readings, anomalies, stats, and health.

3. Real-Time Web Dashboard Built with vanilla JS and Chart.js. Polls the API every 2 seconds and displays live temperature, humidity, and voltage charts alongside an anomaly feed that updates in real time.

4. Three Agentic Test Agents

  • API Test Agent - 36 test cases validating all endpoints
  • UI Test Agent - 27 test cases using Selenium WebDriver
  • Anomaly Detection Agent - 32 test cases validating all 5 anomaly types

All orchestrated through UiPath Automation Cloud as the governance layer.

Challenges I Faced

  • Python PATH corruption on Windows - resolved by fresh install to a user-local path
  • CORS errors - resolved by adding flask-cors explicitly
  • Git push rejection - resolved using --allow-unrelated-histories
  • UiPath Cloud to local API connectivity - UiPath Cloud cannot reach localhost directly, documented and worked around for the demo
  • Bridging embedded and enterprise worlds - required careful architecture design to make C simulator data flow cleanly into the UiPath-governed testing pipeline

Results

✅ 95/95 tests passing across all 3 agents
✅ All 5 IoT anomaly types detected and classified
✅ Real-time dashboard with live sensor visualization
✅ Deployed on UiPath Automation Cloud
✅ Built with Claude Code (AI-assisted development)

Built With

Share this project:

Updates