Parquet

1. DataFrame -> Parquet

DF.to_parquet('{path}/{file_name}.parquet')

2. Parquet -> S3 upload

import awswrangler as wr
wr.s3.to_parquet(DF, path = '{s3_path}', dataset = True)