That's all set up and handy to use locally but I can't see that there's any ClearML support for it. Nor can I think of any way of getting the binary into an instance span up by my auto scaler...

Maybe there's something I can change in the aws autoscaler example