Estimated Time to Complete: 1 hour

In the previous training doc, you saw how the three pieces connect — exporting the model, loading it in the subscriber, and scoring each message. Now it's time to do it for real. You'll run the AI-enhanced subscriber and dashboard, watch normal data flow green, launch the anomaly injector to see orange AI warnings appear, and finally re-run the Project 7 attack simulator to see all three alert colors working together.

Prerequisites

• [ ] Exported model file — Either anomaly_model.joblib from your Colab notebook, or the pre-trained fallback linked here
• [ ] Project 7 dashboard working — Your mTLS broker, certificates, and publisher should still be functional. If they're not running, go back to Launch the Dashboard (Project 7, Step 1) and follow the startup steps again to get everything back online.
• [ ] Project 8 files in place — Download these files and place them in your Project 8 directory:
  • subscriber_dashboard_ai.py
  • dashboard_server_ai.py
  • dashboard_ai.html
  • anomaly_injector.py

Follow the Steps Below

Part 1: Start the AI Dashboard

• Make sure your mTLS broker is running (from Project 5):

  mosquitto -c mosquitto_mtls.conf
  

• Start the AI-enhanced subscriber from the Project 8 directory:

  python subscriber_dashboard_ai.py
  

• You should see:

  ============================================================
  Grand Marina Security Dashboard (AI-Enhanced)
  ============================================================
  AI Model:        anomaly_model.joblib (loaded)
  Dashboard:       <http://localhost:8000>
  ============================================================
  

• The dashboard should open automatically in your browser. It looks like your Project 7 dashboard but with one addition: an "AI Anomalies" counter in the stats footer showing 0.

Part 2: Observe Normal Data

• Start the defended publisher you built in The Defended Publisher (Project 6, Step 2):

  python publisher_defended.py