It's no secret: Microsoft is investing heavily in developing its machine learning capabilities.

Partners have realized this and are increasingly considering the business case for Azure Machine Learning in the context of other Azure services and offerings. MSCloudNews reached out to partners to find out the ways that ML figures into their business plans.

Roshan Pinto heads manufacturing for Microsoft partner Tavant. He explained that data scientists continually refine ML models for tasks as wide ranging as music recommendations and fraud detection. The emerging field of MLOps focuses on the challenges of maintaining, improving, and deploying models over time.

MLOps is a must for enterprises using machine learning at scale. It allows for managing the complete model lifecycle including model governance and should be made mandatory for all Machine Learning projects. Azure Machine Learning provides a great feature set for implementing MLOps. It does lack some of the advanced features like model lineage, but one can always use dedicated MLOps platforms like MLflow or DotScience on Azure to bring in any missing features.

MLOps differs from the similar-sounding DevOps in some critical ways. Non-intuitive mathematical models complicate monitoring, CI/CD involves more moving parts, and coordination is simply different between data scientists and ML modeling researchers.

MLOps also involves different potential risks. Pre-trained models can drift based on data quality, where they are deployed, the degree of built-in scalability, or model governance.

Miranda Yan, founder of Microsoft partner VinPit, sees growing demand and a broadening range of applications for Azure ML capabilities, from churn prediction to sentiment analysis and recommendations. She explained: