Generalize the data fetching in the recommendation workflow from a external URL that changes the data each day.
Change the package pandas_gbq to google-cloud-bigquery to accomplish saving the predictions to google cloud. See https://cloud.google.com/bigquery/docs/pandas-gbq-migration for more information.
Improve the formatting of the recommended movies in Section; Recommendation Workflow.
Go through the CronJob documentation on https://kubernetes.io/docs/concepts/workloads/controllers/cron-jobs/ and run the example cronjob on minikube.
Go through the tutorial on cron by Digitalocean.
Follow the bigquery + python module from https://codelabs.developers.google.com/codelabs/cloud-bigquery-python/index.html#0
Use the managed Airflow solution from Google Cloud, namely Google Cloud Composer to replicate our model training pipeline monitoring.