In many every-day examples, specific constraints hold that prevent optimal performance of machine learning models in the wild. Both their training data and the models themselves are usually subject to many internal and external restrictions. Internal restrictions include the design and maintenance of technological infrastructures, the alignment with business needs, technical debt, or the internal dysfunctionalities of companies. External constraints are related to the accessibility of the data or the legislation companies must obey, among others. Altogether, these restrictions have been studied from different perspectives in the machine learning literature, including accountability, privacy-preserving technologies, fairness, interpretability, data governance, etc., but also from not-so-technical perspectives, such as legal liability, human talent management, firms’ organizational structures, or economic dimensions of machine learning. This Special Issue wants to be a common forum for researchers working on these different machine learning dimensions that must interact in any real-life data centered application.