What does this do?
Reduces the file size of a PDF by re-serializing it with object streams (a more compact PDF structure) and optionally stripping the metadata block. Lossless — the visible pages are unchanged.
How much smaller will my file get?
Depends on the source. A PDF generated by a modern tool may already use object streams and shrink only a few percent. An older or unoptimized PDF (generated by a printer driver or an exporter that wrote everything as separate objects) can shrink 20–40%. Files dominated by embedded JPEG images won't shrink much either way — those bytes are already compressed.
Why don't you recompress the images?
Honest answer: doing it well in the browser requires rendering each page, re-extracting image regions, re-encoding them as lower-quality JPEGs, and splicing them back into the PDF. It's a much bigger pipeline and isn't always lossless. If image recompression is what you actually need, a desktop tool like Adobe Acrobat or Ghostscript will do a better job.
Is my PDF uploaded?
No. Re-serialization happens entirely in your browser.