Machine learning (both training and inference) at scale both for big models as well as massively parallel training and inference of many models

Supporting both individual researchers training and deploying their own models, as well as centralized production workflows using machine learning at scale

Fast inference and training

Calibration/validation

Describe the machine learning training and inference needs of the stakeholders

What are the resources needed to execute these workflows?

How will the stakeholders be able to design and write machine learning applications for these resources

Are there standards that the community should follow?

How are the solutions used by the community embedded/derived from solutions from industry/other science domains