How to normalize JSON into parquet files with foreign keys?

Question

I have a Lambda that receives Events, each containing a nested JSON string which I want to write to S3, in a format queryable by Redshift.

But first, I want to decompose