Architecting MLOps in the Cloud: From Theory to Practice

ML operations (MLOps) refers to a set of practices and tools that automate and combine model development and model operation. MLOps can enable organizations to successfully deploy and manage their ML models in production. The MLOps landscape is increasingly expanding with techno-logical choices, and most public cloud providers offer MLOps platforms with different degrees of maturity. However, this rapidly growing landscape makes designing and implementing an MLOps system in the cloud challenging as the practitioners need to make numerous architectural design decisions, select between different decision options, select between tools/services for realizing a particular decision option, and configure and assemble the chosen tools/services. In this tutorial, we will present guidelines for designing and implementing MLOps for ML-enabled applications in the cloud. We will cover each phase in the MLOps lifecycle. In addition to design guidance, tutorial participants will be able to get hands-on experience in creating an MLOps solution using the Google Cloud Platform (GCP).