Flyte will persist a dataframe (parquet) on remote storage such as S3 by passing the

uri

param.

def load_data() -> StructuredDataset:
    return StructuredDataset(dataframe=pd.read_csv("..."), uri="<s3://my-bucket/my-dataframe.parquet>")

However, this results in partitions nested under

my-dataframe.parquet

. Is there a way to directly save without the partition structure.

|- my-bucket
|  |- my-dataframe.parquet
|  |  |- 0000

0000 is default file name, flytekit didn’t create partition by default