But, to be fair, they’re not dedicated to only keeping the lights on, they probably spend 20% (1 eng year per year) of their collective hours fighting fires, so I don’t think burnout is too much of an issue. The remainder of their time is spent maintaining libraries and web interfaces. They’re a pretty standard “platform engineering” team.
In my perspective, it doesn’t matter what technology you go with (K8s or otherwise): a responsibly-managed your platform team requires at least one team with a full on-call rotation (i.e. at least 4 engineers), depending on how wide your golden path is.
Honestly curious - never used k8s or Mesos* in production and we have a much smaller team managing a much smaller infrastructure, I am sure.
edit: meant to say mesos but applies to both