This paper describes an adaptive approach to developing machine learning operations platforms that incorporates goal-oriented requirement engineering using an organizational modeling framework, microservice architecture principles, and small language models. We demonstrate this method with a case study of a machine learning workflow management application. The proposed model enables more efficient integration of artificial intelligence capabilities while maintaining system modularity, scalability, and alignment with stakeholder needs and expectations. Our results show that the goal-oriented MLOps approach for microservices-based systems provides considerable benefits in managing complex machine learning workflows, particularly when incorporating several artificial intelligence services with different features and requirements.