This is a very innovative take on infra for ML observability. Shreya Shankar and collaborators at Berkeley came up with https://github.com/loglabs/mltrace, which treads the same ground - perhaps you've looked at it already?
Thanks! And thanks for sharing the pointer - I think I've seen `mltrace` at some point in the past. The tool has some similarities, but seems different from `mandala` on a philosophical level - `mltrace` seems more opinionated and domain-specific to ML. `mandala`'s goal is to make persistence logic a more generic part of Python semantics, and there's much more emphasis on composing `@op`s freely in whatever complex ways you want. Python functions are great and everyone knows what they do - let's give them more superpowers!
If this is something you're interested in, I can probably give a more detailed comparison if I find the time.
