See inside your Parquet files.
Explore schema, stats, and rows in a fast data grid. Open local or remote Parquet, CSV, TSV, JSON or JSONL — private, never uploaded.
Explore a public dataset in one click.
Each card opens a public dataset straight from its https:// URL. Parquet streams in place; CSV, TSV and JSONL download once, then open locally.
Iris
Parquet · 5 KB150 flowers with four petal & sepal measurements and the species of each.
Petal & sepal averages per species Hugging FaceTitanic
Parquet · 17 KBPassenger manifest with ages, fares, cabins and who survived.
Survival rate by class & sex Hugging FaceHeart disease
Parquet · 8 KB261 patients with age, chest-pain type, blood pressure and diagnosis.
Diagnosed patients, oldest first Hugging FaceNYC Airbnb
Parquet · 3.7 MB49K listings — prices, neighborhoods, room types & review dates.
Priciest Manhattan listings Hugging FaceCensus income
Parquet · 540 KB33K people with age, job, education, hours worked and income bracket.
High-income share by education Hugging FacePalmer penguins
CSV · 13 KB344 penguins with species, island, body mass and bill & flipper measurements.
Heaviest penguins first GitHubUS unemployment
TSV · 34 KBUnemployment rate for 3,218 US counties, keyed by county FIPS code.
Highest unemployment first GitHubDolly 15k
JSONL · 13 MB15,010 instruction, context & response records — a classic LLM fine-tuning set.
Creative-writing prompts only Hugging FaceAny common format, wherever it lives.
Parquet first, but it reads CSV, TSV, JSON and JSONL too — from disk, a URL, Hugging Face or S3.
Formats
Sources
Everything, at a glance.
Columnar storage, a columnar viewer. Types are color-coded, numbers are tabular, metadata is one glance away.
Read in place, never uploaded
Files are read where they live — only the bytes your view needs are streamed. Nothing hits a server, so even multi-gigabyte files open in moments.
Schema & statistics
Walk the full schema — types, nesting, repetition — each column color-tagged. Min, max, distinct count, null ratio and size come straight from metadata. No full scan.
Filter, sort & query — or ask AI
Filter and sort with type-aware operators, or run full SQL — joins, aggregates, CTEs, windows. Or describe what you want instead, and let AI write the SQL.
Export & share
Export the current view to CSV, JSON or Parquet, written locally. Or share a link to exactly what you see — it rides the URL #hash, so it never reaches our servers.
From file to slice in about a minute.
Open anything
Drag in files or a folder — any format, even
mixed. Or point at a remote URL,
hf:// dataset, or
s3:// bucket.
Inspect & filter
Read the schema and stats, then filter and sort with type-aware operators — or describe a query and let AI write the SQL.
Export a slice
Export the current view to CSV, JSON or a new Parquet file — written straight to a folder you choose.
Two ways to open a file.
Same private, columnar viewer — run it in your browser, or install the native macOS app.
Web viewer
Web app · nothing to installOpen the viewer in any modern browser. Files are read locally — never uploaded.
Open web viewer Chrome · Safari · Firefox · EdgemacOS app
Native · works fully offlineA native build with folder access, drag & drop and faster reads on very large files.
Download on the App Store Apple silicon & Intel · macOS 15+