SageMaker Connector

Route device telemetry to Amazon SageMaker inference endpoints for real-time machine learning predictions. Enable anomaly detection, predictive maintenance, and quality classification on your live IoT data stream.

What is SageMaker

Amazon SageMaker is a fully managed machine learning service that enables data scientists and developers to build, train, and deploy ML models. The /IOTCONNECT™ connector routes telemetry to SageMaker real-time inference endpoints for instant predictions.

Key Features:

Real-time Inference: Get ML predictions on live telemetry data within milliseconds.

Multi-framework Support: Deploy models built with PyTorch, TensorFlow, XGBoost, scikit-learn, or custom containers.

Auto-scaling Endpoints: Inference endpoints scale automatically based on request volume.

A/B Testing: Run multiple model versions simultaneously to compare performance.

Model Monitoring: Detect data drift and model degradation with built-in monitoring.

How it Works

When the SageMaker connector is enabled, /IOTCONNECT™ sends device telemetry to your SageMaker endpoint and returns predictions:

Device sends telemetry → Sensor readings arrive at /IOTCONNECT™.

Data prepared → Telemetry is formatted as input features matching your model’s expected schema.

Inference request → Data is sent to your SageMaker real-time endpoint.

Prediction returned → The model returns predictions (anomaly score, classification, regression value).

Action triggered → Predictions can trigger alerts, rules, or downstream automation.

Video Walkthrough: Amazon SageMaker

Watch the complete walkthrough video showing the end-to-end process of configuring Amazon SageMaker and integrating it with the IoTConnect platform for AI model deployment and real-time device telemetry processing.

User Manual

Step 1: Navigate to Connector Settings

Go to Settings > Configuration > Live Data Streaming in the /IOTCONNECT™ portal.

Step 2: Enable SageMaker

Find the SageMaker connector card and toggle the switch to ON (green).

Step 3: Select Data Types

Choose: Telemetry (primary for ML predictions).

Step 4: Run CloudFormation Stack

Click the CloudFormation link. This creates:

IAM cross-account role for SageMaker endpoint invocation

Required IAM policies (sagemaker:InvokeEndpoint)

Step 5: Configure Endpoint

Provide your SageMaker endpoint name. The endpoint must be deployed and active in your account.

Step 6: Save

Click Save. Telemetry begins flowing to your ML model for real-time inference.

Use Cases

Anomaly Detection: Detect unusual sensor readings indicating equipment failure or environmental hazards.

Predictive Maintenance: Predict when equipment will fail based on vibration, temperature, and usage patterns.

Quality Classification: Classify product quality in real time on manufacturing lines.

Energy Optimization: Predict optimal HVAC settings based on occupancy and weather data from IoT sensors.

Demand Forecasting: Forecast resource consumption patterns from smart meter data.

ⓘ NOTE

Your SageMaker endpoint must be deployed and active before enabling this connector. Endpoint costs accrue based on instance type and uptime. Use auto-scaling to optimize costs for variable IoT workloads.