We'll need to change the data pipeline output format to make easier to load & manage in Cassandra.
Two things that come to mind :

• Fix. unescaped " characters
• image_id should have default, non-null value

More details might come up once we finalise the schema.