Methodology based on MLOps (Machine Learning Operations) for management support in data science projects

Abstract

: Many engineering companies have made the strategic decision to venture into the field of data science and MLOps in order to create, extract, and analyze vast amounts of data. This approach holds significant importance due to its inherently multidisciplinary nature, combining principles, concepts, and practices from various domains including engineering, machine learning, and mathematics. The primary objective of exploring this field of work lies in achieving high performance, efficiency, and effectiveness by correctly applying the concepts, methodologies, procedures, and guidelines offered by this area of study. However, it is imperative to acknowledge that despite the clarity of the definition and concept, there remains a dearth of information concerning the specific methodologies and procedures required for executing projects of this nature. This scarcity can be attributed to the relative novelty of the term MLOps. Therefore, this paper presents a comprehensive methodology based on MLOps that serves to facilitate data science project management. It is worth mentioning that while this project is grounded in the context of a Colombian company, extensive research has been conducted, encompassing various countries through the exploration of relevant literature, papers, documents, and information from diverse sources