Create a Argo Workflow for MLOps

[ebrucucen]

We want a repeatable and reliable process for data ingestion. The options we had were:

• Kubeflow (YAML based)

• Cloud Composer ( use Apache Airflow, Python-based)

• Apache Beam (Python Based)
  We decided to have ArgoCD as it is simple to define the tasks and reference with a simple DSL.
  We will need

• A GKE to be deployed to GCP

• ArgoCD should be deployed to this GKE

• At least one container to call from the workflow

[ebrucucen]

[] We have deployed GKE (with default settings, may need to readjust the settings tomorrow to check the consumption vs cost)
[] We have deployed ArgoCD following the documentation:
https://argocd-operator-helm.readthedocs.io/en/latest/k8s/gcp.html
(https endpoint throws an error, and redirect to non-https endpoint... we need to check it out)