Productionizing Your Workload with MLOps
MLOps is a concept that enables machine learning (ML) workloads to scale through the automation of model training, model evaluation, and model deployment. MLOps enables traceability with code, data, and models. MLOps allows data scientists and ML professionals to make predictions available to business users at scale with the Azure Machine Learning (AML) services.
MLOps is built on the concepts of CI/CD. CI/CD is a term that stands for continuous integration/ continuous delivery and has been used for software development for decades. CI/CD enables companies to scale their applications and by leveraging those same concepts, we can scale our ML projects, which will rely on CI/CD practices for our MLOps implementation.
One of the challenges of this domain is its complexity. In this chapter, we will go through the scenario of retrieving data, transforming data, building a model, evaluating the model, deploying a model, then pending approval...
