Hi Bob,
Indeed, I have found there is a large gap between hello world examples and doing it in real life. As such, I’ve tried to create a youtube channel dedicated to teaching Stan for absolute beginners (Learn Stan with Ric - YouTube).
In my opinion, teaching the Math Bayesian Inference and learning Stan should not be independent exercises. Practical Bayesian inference should be learnt hand in hand with Code, and also vice versa In the past when someone has asked how to learn Stan, I had told them to learn from the examples in the User’s guide. While the examples are good, those without a solid fundamental grasp of what is happening tend to get stuck quite early. So, I created this playlist (Bayesian Inference with Stan - YouTube) that goes through fundamentals and examples.
Indeed, I think it’s a good idea to focus on Python. Stan is an amazing language, but there is definitely a perception that it is more a “research” tool, than a tool that can be used in production. I know this isn’t the case (I have put Stan models into full production before using cmdstanpy), but having more tutorials out there using Python as the interface is exactly what we need.