Spark is great for data engineers. Data scientist should not really be on Spark. Databricks though is catching up with a new distributed Python Pandas API no need for Koalas. Could you compare this Saturncloud to Azure ML? Like your articles:)