Performing machine learning and operations (MLOps) on a single computer can be challenging. When we think about training, deploying, and maintaining models across thousands of computers, the complexity of doing so can be daunting. Luckily, there are ways of reducing this complexity using tools such as containerization and continuous integration/continuous deployment (CI/CD) pipelines. In this chapter, we are going to discuss deploying models in a way that is secure, updatable, and optimized for the hardware at hand.
In terms of building updatable models, we are going to discuss using Azure IoT Hub Edge devices to enable over-the-air (OTA) updates across a single management plane. We are also going to use device twins to maintain the fleet and push configuration settings going to our models. In addition, we'll learn how to train a model on one type...
