Comment by wenc

Yes. My job involves pulling a ton of data off Redshift into Parquet files, and then working with them using DuckDB (sooo much faster — DuckDB is parallelized, vectorized and just plain fast on Parquet datasets)