After switching from autopilot cluster to a standard cluster, autoscaling was set up.

We should evaluate the cost of a standard cluster with autoscaling. It might not be well optimized for running a single nightly job.

Another option is resizing the cluster inside the nightly job script. 

scale up -> run the job -> scale down

See 

• https://cloud.google.com/sdk/gcloud/reference/container/clusters/resize 
• https://bhabalajinkya.medium.com/scaling-down-google-kubernetes-cluster-to-zero-during-off-peak-hours-4281b464d14e