As with many other areas of science, the quantity and quality of available cosmological data has exploded in recent years. This increase undoubtedly presents tremendous scientific opportunity. But accessing and analyzing this data at scale presents real challenges, especially for scientific users who are not familiar with HPC or big-data workloads.

The OpenCosmo project exists to build a domain-specific abstraction layer on top of the cosmological datasets and computing resources housed at various DoE facilities. It aims to allow scientific users to describe which data they want and what analysis they would like to perform, without knowing or caring which facility the work is being performed at. The platform is a specific example of a more general trend towards "Science as a Service" tools that are cropping up in many different domains. Understanding the shared patterns and infrastructure needs of these tools is essential to ensure their viability and continued success.