awswrangler returning dataframes with different dtypes

Question

I am using awswrangler to convert a simple dataframe to parquet push it to an s3 bucket and then read it again. This is the code:

import boto3
import aw