The solution isolates workloads
by hardware-based memory encryption within public or private cloud infrastructures so that no one, including cloud providers,
ML model developers, system administrators or any other third parties, can access private data
or trained models.