In lieu of official hosting which was requested a year ago, go ahead and refactor approach to work on AWS S3. The previous approaches used either soft or hard links to avoid duplicating files that remain unchanged between snapshots, but S3 is not a file-system and does not support links. The closest comparable concept are object level redirects, but managing and creating those would be combersome, slow, and result in a higher bill. To avoid that an alternative that does not require object level redirects to simulate links, but that does not requiring duplicating files would be preferable.
In addition the tool built to provide such snapshots on S3 should be easily deployable and setup to run automatically on AWS.
This project is one of its kind!