Throughout this book, we've learned all about how to create Artificial Intelligence (AI) applications to perform a variety of tasks. While writing these applications has been a considerable feat in itself, it's often only a small portion of what it takes to turn your model into a serviceable production system. For many practitioners, the workflow for deep learning models often ends at the validation stage. You've created a network that performs extremely well; We're done, right?
It's becoming increasingly common for data scientists and machine learning engineers to handle their applications from the discovery to deployment stages. According to Google, more than 60-70% of the time it takes to build an AI application is spent on the deployment architecture of that application. Given that this book is designed to...