I really like Vertica's export to Parquet feature and also the incredible speed at which it can import Parquet files directly from an AWS S3 bucket.
It's also no slouch at importing JSON attributes from files directly into columns provided those column names reflect the JSON path for the attribute.
Under the hood it uses Apache Arrow for translating its columnar format to other columnar formats.
It uses a C library called YAJL to import JSON.
I'm very impressed with Snowflake and the way it uses integrations and stages.
It has its own task scheduler that you setup using a SQL CREATE TASK command. In fact everything seems to have a well thought out SQL command to support it.
The thing that is most impressive is how comprehensive the SQL implementation is, especially considering how relatively young Snowflake is