MLOps: A Guide to its Adoption in the Context of Responsible AI

Abstract

DevOps practices have increasingly been applied to software development as well as the machine learning lifecycle, in a process known as MLOps. Currently, many professionals have written about this topic, but still few results can be found in the academic and scientific literature on MLOps and how to to implement it effectively. Considering aspects of responsible AI, this number is even lower, opening up a field of research with many possibilities. This article presents five steps to guide the understanding and adoption of MLOps in the context of responsible AI. The study aims to serve as a reference guide for all those who wish to learn more about the topic and intend to implement MLOps practices to develop their systems, following responsible AI principles.CCS CONCEPTS• Software and its engineering → Software creation and management; • Computing methodologies → Machine learning.