I haven't seen information about native support in Mathematica for reading Apache Parquet files. I had the same question myself this week, but could not see much information on the internet. I successfully read a parquet file using Python, then imported the result into Mathematica as follows:
session = StartExternalSession["Python"]
ExternalEvaluate[session, "import pandas as pd"]
ExternalEvaluate[session, "df = pd.read_parquet('C:/test.parquet')"]
df = ExternalValue[session, "df"]
Mathematica imports the pandas dataframe as a Dataset. I hope this helps!
