It's completely overkill for pretty much everyone but I have been thinking about building a kubernetes native client for months now.

Like the torrent should be treated as a normal resource with a Torrent CRD. It should be scheduled onto whichever node has available capacity and rescheduled onto a different node if it goes down. If allowed by the tracker, multiple instances could be run. You could set resource limits programmatically, easily configure block storage, build dashboards, export logs/metrics.. It would be open ended enough that you could have interfaces built as browser extensions, web ui, mobile app, tui, cli and be unopinionated so much that the method for torrent ingestions could be left up to the used. HTTP request, watch directory, rss client, download manager.. You could even do stuff like throw magnet links into a queue.. etc, etc..

I keep thinking it would be a great project but I just do not have the spare time to dedicate to it.. I imagine it could be used for large scale deployments for something like the Internet archive or whatever.