We just released cuthbert , a modular python package for working with time series - specifically probabilistic state-space-models, think Kalman or particles filtering and smoothing.

We put a lot of thought into the code structure to ensure it supports (temporal) parallelisation, composition with JAX ecosystem and can be easily extended with new methods.

Would love feedback, issues, or ideas for where it should grow next! (Just don't say into a butterfly)