ViewKit

ViewKit is a browser-based data file viewer built for machine learning researchers and data scientists who need to inspect datasets without uploading them anywhere. The entire application runs client-side, which means your data never leaves your machine. You open a file, it processes in your browser, and that's it. No server, no upload, no third party ever sees your work.

The problem it solves is one that anyone working with research data knows well. You have an HDF5 file or a Parquet dataset sitting on your machine, and you just want to look at it. The traditional path is to fire up a Jupyter notebook, import the right libraries, and write a few lines of code to peek at the structure. ViewKit skips all of that. You drag the file into the browser, and you're looking at the data in seconds.

Format support is broad and clearly aimed at the ML ecosystem. It handles HDF5, Parquet, Zarr in both v2 and v3, CSV, TSV, Excel, JSON Lines, Apache Arrow, Feather, NumPy arrays in npy and npz form, safetensors, TFRecord, WebDataset tar archives, and a range of audio files including WAV, MP3, FLAC, OGG, and M4A. That coverage means you can open most of the common artifacts from a training pipeline without switching tools or installing anything locally.

The audience is anyone who touches datasets in a research or production ML context. If you're debugging a data pipeline, checking the shape of tensors before a training run, or just exploring a dataset you received from a collaborator, ViewKit lets you do that in the browser. There's no account to create, no installation to manage, and no data leaving your control.

What sets it apart is the combination of breadth and simplicity. Plenty of tools can open a CSV, but fewer handle the more specialized formats like safetensors or TFRecord without requiring local setup. And because everything runs in the browser with client-side processing, you can use it on a locked-down work machine where you can't install arbitrary software. The privacy story is strong because there's nothing to trust beyond your own browser.

ViewKit is free with no paid tier mentioned. You visit the site, open your file, and inspect your data. There's light and dark mode support that persists between sessions, but the core value is the instant, zero-friction inspection of datasets in a format-aware viewer that never phones home.