File Format	Source Data Type	Destination Data Type	Notes
CSV	Inferred from data	string	All CSV fields initially treated as strings
CSV	Numeric patterns	int, bigint, double	Integer and floating-point numbers auto-detected
CSV	ISO 8601 dates	timestamptz	Date/datetime strings converted to timestamp
CSV	Boolean values	boolean	true/false strings converted to boolean
JSON	string	string	JSON string fields
JSON	number (integer)	bigint	JSON integer values
JSON	number (float)	double	JSON floating-point values
JSON	boolean	boolean	JSON boolean values
JSON	object, array	string	Nested objects/arrays serialized to JSON strings
JSON	null	string	Null values converted to empty strings
Parquet	STRING, BINARY	string	Parquet string types
Parquet	INT32, INT64	int, bigint	Parquet integer types
Parquet	FLOAT, DOUBLE	float, double	Parquet floating-point types
Parquet	BOOLEAN	boolean	Parquet boolean type
Parquet	TIMESTAMP_MILLIS	timestamptz	Parquet timestamp types
Parquet	DATE	date	Parquet date type
Parquet	DECIMAL	float	Parquet decimal types converted to float64
All Formats	_last_modified_time	timestamptz	S3 LastModified metadata (added by connector)

Schema Inference

• CSV: Uses AND logic - examines all sampled rows to determine most restrictive type
• JSON: Auto-detects types from JSON primitives
• Parquet: Schema read directly from file metadata (no inference needed)

timestamptz timezone

OLake always ingests timestamp data in UTC format, independent of the source timezone.