Data Science Analytics and Artificial Intelligence/Machine Learning (DSA-AI/ML) applications aim to create decision-making value by applying analytical procedures from Computer Science, Statistics and AI/ML disciplines to small or big internal and external datasets. However, the development and utilization of applications in this domain requires an adequate DSA-AI/ML architecture. Hence, whereas there is a varied academic literature on DSA-AI/ML Reference Architectures (RAs)—i.e. generic high-level architectural templates for implementing a specific architecture -, such are scarce studies from a dual ISO/IEC/IEEE 42010 Architectural Standard and IT Service Design approach. The ISO/IEC/IEEE 42010 Standard provides valuable guidance for documenting architecture designs and the IT Service Design approach provides best practices to deliver cost-effective IT services. In this chapter, thus, we address this knowledge gap and provide a novel selective-integrative review of main DSA-AI/ML Reference Architectures from a dual ISO/IEC/IEEE 42010 and IT Service Design approach, in the context of non-proprietary domain-independent RAs. Our descriptive review provides theoretical—an updated DSA-AI/ML Architectural Framework—and practical insights—brief but useful descriptions on the purpose and technologies for the DSA-AI/ML Architectural Framework—useful for understanding DSA-AI/ML Reference Architectures.