Building AI applications goes beyond the basics of model construction – it takes deploying your models to a production environment in the cloud where they persist.In this chapter, we've discussed how to take a validated TensorFlow model and deploy it to production in the cloud. We also discussed ways that you can scale these models, and how you can test your applications for resiliency.
When taking a TensorFlow application from development to production, the first step is to create a TensorFlow SavedModel that can be stored in the cloud. From here, there are several services, including AWS Lambda and Google Cloud ML Engine, that can help make your deployment process easily.
Applications can be scaled up or out for more computing power and faster processing. By scaling up, we provide our algorithms with a larger computing resource. By scaling out, we provide...
