Is MLOps basically Data Engineering rebranded?

Abstract

In this talk, we will review what MLOps entails, why DevOps alone wasn’t sufficient, and whether MLOps can be viewed as a mere rebranding of Data Engineering. By combining the principles of DevOps with specialized techniques tailored to machine learning, MLOps aims to provide a streamlined and efficient approach to deploying and managing machine learning models in production environments. However, some experts argue that MLOps is simply a repackaging of data engineering - a discipline that has been around for several years. They contend that many of the fundamental principles of MLOps, such as data preparation, data quality assurance, and model deployment, have long been part of the data engineering toolkit. In this talk, we will examine these viewpoints in-depth and try to gauge the validity of each.