what are those?
it appears to be a process created by the CUDA runtime in some situations in order to handle internal housekeeping tasks
I don’t think there is any specification about it, and I don’t think there is anything you can do about it.
Any expectation that CUDA should use zero system resources is mistaken.
Any expectation that CUDA behavior is static and will never change is mistaken.
The fact that certain aspects of CUDA are unpublished is often for a reason. If there is no public specification, then there is no reason to base your expectations on unpublished information.
If you’d like to see a change in any aspect of CUDA, you can file a bug. The instructions to do so are linked to a sticky post in this sub-forum.