Take the components of the Jupyter notebook and convert them into a more traditional Python Object/Class. Have the Class save the dataframe to parquet and push to S3.