privacy

csvfit is a static site that runs entirely in your browser. There is no backend and no account system. This page describes exactly what happens to your data — which is almost nothing.

Your CSV data

When you drop a CSV onto the page, the file is read into browser memory and handed to a Web Worker running Pyodide. Training, evaluation, and feature-effect computation all happen inside that worker, in your tab, on your CPU. The data is never transmitted over the network. Closing the tab discards it.

Share links encode the model and a small subset of the dataset into the URL fragment (the part after #). Browsers do not send URL fragments to servers, so share links are decoded locally in whoever opens them — the CDN only sees the request for the page itself.

Analytics & tracking

None. No Google Analytics, no Plausible, no heatmaps, no session replay, no third-party scripts. Open DevTools > Network and train a model — you will see the initial page assets, the Pyodide runtime, and a Python package snapshot. Nothing is sent back.

Cookies & storage

No cookies are set. The site uses localStorage to remember one preference — whether you have opted into auto-loading the sample dataset on boot — under the key fastml:autoload. That value never leaves your browser.

Hosting

The static site is served from Cloudflare Pages. Cloudflare sees the IP address of the request and the path requested, as is the case with any website. No server-side logic runs on your CSV or model.

Contact

Questions or concerns: