Client-side processing is the right call for PDFs, since people routinely run contracts and IDs through these tools. Curious how you handle very large files in memory. We've been tackling the same problem with a no-signup compressor at ToolSphare: https://toolsphare.com/tools/compress-pdf