Since the topic is about user experience, I hope my comments are relevant.
I spend most of time writing ODEs systems for infectious disease modelling. I appreciate that there are several research papers out there that provided the workflow for the analysis based on STAN (especially after COVID). Those workflows are quite useful and relevant to me too. At the time when I started modelling though, I found the model specification in STAN a bit daunting because my model structures are quite hard to formalise into equations into STAN. If those workflows were available at the beginning, I probably would have made a different decisions.
PyMC was easier for me to pick up. The process from ideas to model structures to codes is faster too. I think because I was more familiar with Python so that’s that.
On the side note, I saw this tutorial for Numpyro in epidemiology. Look very interesting. After I finished my thesis, I will have look at them. My feelings is Numpyro is also very Python-centric (?)