pymc-devs/pymc ? reverse-engineered prompt

Build me a Python package for Bayesian statistical modeling like PyMC.

I want to be able to define simple probabilistic models in a clean, readable way, run modern sampling methods for inference, and also have a fast approximate option for when I need results quickly. It should support things like normal regression models, latent variables, missing data handling, prior and posterior predictive sampling, and summarizing the inferred parameters.

Please make the experience feel friendly for someone who knows a little statistics but does not want to build all the math machinery by hand. A small example app or demo notebook showing linear regression on fake data would be great, along with clear docs or comments so I can understand how to use it.

If you need to check current docs or best practices online, go ahead.