There is support for running on GPU which will be beneficial to certain ML workloads.Clusters are configured to run on CPU by default, but you can choose GPU based nodes during creation.
I'll add that you may wish to hit the endpoint more frequently if your web service can take it... it's wasteful to poll once every five minutes for a response that might return at 5 minutes and 1 second.
Nothing wrong with using Sleep per se, but this business logic lacks timeout logic.If it never returns success, your job will run forever, consuming DBUs as well as cloud compute costs.Using the scala concurrent duration classes, you can create a...
