We deal with fairly large volumes of data on a frequent basis so it would not make sense for each data scientist to create a copy within their own environment. Everyone works off a centralized data source and we provide them with Jupyter/Spark in an internal cloud environment.