Schemas — auto-detection and overrides

Data Stream automatically infers a schema from the structure of your data on the first job. Schemas are displayed and managed in the Schema tab.

What a schema contains:
• Field keys — the raw property names from the source data
• Labels — human-readable display names (editable)
• Types — string, number, boolean, date, array, or object
• Required — whether the field must be present in every record

Auto-detection:
• The first time you process a new type of data, a schema is generated from the output and saved automatically
• The schema name matches the source job it came from

Overriding a schema:
1. Go to Data Stream → Schema
2. Click on a schema to expand it
3. Edit the label, type, or required flag for any field
4. Click "Save Override" — the schema is now user-controlled and will not be overwritten by future auto-detection

When to use schema overrides:
• Your source data has field names that are hard to read (e.g. "f1", "col_002") — set readable labels
• You want to enforce specific types (e.g. always treat "amount" as a number even when it arrives as a string)
• You process the same dataset repeatedly and want consistent output structure

Deleting a schema:
• Click "Delete" on any schema row — this removes the saved schema but does not affect jobs already run with it