Build and deploy an end-to-end recommendation system architecture with Cloud Composer and Apache Airflow using Cloud Machine Learning Engine to retrain the model, and then sending a job to App Engine to redeploy the updated application endpoint.
In this talk, we will review a few end-to-end recommendation models architectures and we will learn how to implement an ML data refresh workflow to periodically retrain and redeploy a recommendation model in order to serve fresh news articles data via an API endpoint. We will also learn how to use Cloud ML Engine for submitting Tensorflow training jobs, and App Engine where we'll deploy and redeploy our model for serving while keeping up to date with fresh recommendations. For our data pipeline, we will set up Cloud Composer and Apache Airflow environments to be used as the overall orchestration layer to schedule, call and send tasks to other GCP services such as BigQuery and Google Cloud Storage.