I was asked to explain etcd, controllers, kube-apiserver, basically the entirety of control plane, kubelet, kube-proxy, ingress, services, load-balancer and how all of them work with each other. For architecture, I told that I can create taints and tolerations to seperate pods for frontend on seperate nodes and pods for back-end on seperate nodes. I did say that there can be better ways of doing this provided more info, but they said ok.