They talk about scientific workflows but don't mention spack.

That is a valid criticism. Although I personally have seen Spack come up mostly in HPC workloads which is a bit more specialized. There are many disciplines that might not need the "full power" of something like Spack (e.g. source builds) but just need to setup some computational tooling for computational biology research, physics, etc.

See also:

- Gentoo Prefix https://wiki.gentoo.org/wiki/Project:Prefix

- European Environment for Scientific Software Installations (EESSI) https://www.eessi-hpc.com/

- REANA (Reusable Analyses) https://reanahub.io/

How does this compare with rye?

There are definitely some similarities, but pixi is explicitly not "Python-only". We support many more package ecosystems (thanks to Conda being language agnostic).

This includes, for example, R, C/C++ compilers, Rust, Go ...

I think it's more comparable to Nix or Docker, actually.