Convert your Parquet files to JSON format for APIs, web apps, and data interchange. Files are processed locally in your browser.
How the Parquet to JSON conversion works
Drop a .parquet file, preview the rows, and download a .json containing an array of row objects — one per row, keyed by column name, pretty-printed. Decoding runs locally in your browser; nothing is uploaded. The records orientation loads directly with JSON.parse, Python's json module, or jq.
How Parquet types map to JSON
BOOLEAN → true / false
INT32 / FLOAT / DOUBLE → number
INT64 → string (JSON numbers are floats and would corrupt large integers)
TIMESTAMP → ISO 8601 string
BYTE_ARRAY / UTF8 → string
NULL → null
Need newline-delimited output? Use the NDJSON converter — same values, one object per line.
Large files and limits
Reads cap at the first 100,000 rows; pretty-printed JSON is verbose, so a full export can reach tens of MB. Bigger jobs: pandas to_json(orient='records') or DuckDB COPY … (FORMAT JSON, ARRAY true).