Is there a way of creating a gitlab-runner-like interface for dagster where we can configure a server as a executor?

Not that I can think of. What are you trying to do with this? I'll say that you can likely make an abstraction/interface by using our Python APIs to programmatically trigger the webserver, and configure your executor then.

I was just wondering if there is a binary we could throw on a machine to for example use something like an Ec2 autoscaling cluster (without first having to attach it to EKS, taint + label, etc.). Or in our case, we have 1 special GPU machine for experiments that we would like to use. I understand that we could always ssh into that machine (paramiko or otherwise, I just saw the dagster executors as promising to keep the same interface as well as abstract out some setup/teardown + queueing logic that'd be common among all executors