Explainable MLOps: A Methodological Framework for the Development of Explainable AI in Practice

The field of explainable artificial intelligence (XAI) aims to increase the transparency of AI models by providing explanations for their reasoning processes. Valuable efforts have led to an increase in transparency. However, there are still blind spots in literature, specifically related to the use of XAI techniques in practice. To make the development of AI models truly explainable, transparency is required in each stage of the Machine Learning Operations (MLOps) workflow: data preparation, model development and model deployment. This research aims to mitigate issues in each stage, using case studies from the industry and health domains. The final objective is to provide an application-oriented methodological framework for the development of more transparent AI approaches.