Parquet is great for pipelines and terrible for humans. Drop a .parquet file and actually see the rows — sort, search, then export to CSV or Excel if you need it elsewhere.
Normally 'just check the Parquet file' means spinning up pandas or DuckDB. Here the decoder runs in your browser: drop the file, see the rows.
The Parquet decoder (a small pure-JS reader) lazy-loads the first time you use it — no WASM, no cluster.
Column chunks are decoded and assembled into ordinary rows. Snappy, gzip, zstd, brotli and lz4 compression are all supported.
Sort, search and copy like any grid — or export the data as CSV / Excel, or open it in the dashboard builder.
The quickest way to answer "what's actually in this file?"
No pip install, no JVM, no notebook. A browser tab is the whole toolchain.
Snappy, gzip, zstd, brotli and lz4 page compression decode in-browser.
INT64, timestamps, decimals, byte arrays and nested structures all render readably (nested values as JSON).
Data-lake files often hold sensitive data — this never leaves your machine.