🗄️ SQL Data Workbench
Run SQL queries on CSV, Parquet, and JSON files with JOINs between datasets
Drop CSV, JSON, or Parquet files here or click to browse
Up to 5 files, 50 MB each
Loading DuckDB engine…
About this tool
SQL Data Workbench lets you run SQL queries across CSV, JSON, and Parquet files — including JOINs between different datasets — without installing anything. Powered by DuckDB's embedded analytical engine running in your browser.
Drop in a sales.csv and customers.parquet, then run SELECT c.name, SUM(s.amount) FROM read_csv_auto('sales.csv') s JOIN read_parquet('customers.parquet') c ON s.cust_id = c.id GROUP BY c.name to get per-customer revenue totals. Results appear in a table and can be exported as CSV or JSON.
The engine supports standard SQL: SELECT, JOIN, GROUP BY, window functions, CTEs, subqueries, and more. DuckDB auto-detects CSV delimiters and column types so you can focus on the query.
FAQ
What file formats are supported? ›
CSV (including TSV and other delimiters), JSON (arrays of objects or newline-delimited JSON), and Parquet. DuckDB auto-detects the format from file extensions and contents.
Can I join data between different file types? ›
Yes. DuckDB treats all files as queryable tables using read_* functions. You can JOIN a CSV file with a Parquet file, or UNION JSON with CSV — the engine handles type coercion and schema matching automatically.
Where does my data go? Is it uploaded anywhere? ›
No. Files stay in your browser's memory — DuckDB runs inside a WebAssembly worker, and the engine queries registered file buffers directly. Nothing leaves your machine.
What SQL features are available? ›
DuckDB supports standard SQL including SELECT, WHERE, GROUP BY, HAVING, ORDER BY, JOINs (inner, left, right, cross), window functions, CTEs (WITH), subqueries, CASE expressions, and aggregate functions (SUM, AVG, COUNT, MIN, MAX, etc.).