My understanding is that Kubernetes is more efficient UI for managing large clusters of containers, otherwise you're stuck with the command line. As of August 2018, It seems multi-platform support on Docker for Windows is still experimental (https://github.com/moby/moby/issues/33850, https://github.com/docker/for-win/issues/2079). Running any Linux image with the --isolation hyperv after pulling with the --platform linux switches doesn't work if not running on expermintal flag. How to enable to experimental flag on Docker for Windows Server core isn't documented. So setting up a single node kubernetes cluster on Windows Server 1803+ for QA purposes probably isn't well supported or even documented. Being an exclusive Windows shop, having to use a Kubernetes on Linux doesn't seem to make sense especially when the whole point of using Docker is to automate environments. What's the point of making environment configuration automatic when you still need to administer a Linux server.
How do you setup Kubernetes on Windows Server Core 1803 as a single node cluster?
You can quite easily set up one node Kubernetes cluster on Windows using minikube. You can use a native hypervisor (Hyper-V) if you want, but it is more recommended to use VirtualBox at this time.
In both cases, a Linux virtual machine will run on your server, and you will be able to access it with the native version of kubectl to manage the cluster.
Here are two manuals that can help you to do that (I can guess it will work on the Windows Server in the same way):
Here is an explanation of how you can use Windows server as a worker node in Kubernetes:
Microsoft announces that the next version of Windows Server will have better support of Kubenetes and it's ready for testing: