Querying Parquet directly

You don't import Parquet — you query it where it lies:

SELECT status, COUNT(*)
FROM 'events/*.parquet'
GROUP BY status;

Glob patterns scan a directory as one table; DESCRIBE shows the schema that travels inside the file:

DESCRIBE SELECT * FROM 'events/2026-01.parquet';

Files as tables, schema included, no load step — this is why Parquet is the lingua franca between data tools.