I haven't used pymc3, but Stan is a state-of-the-art programming language for Bayesian inference and has interfaces to R, Python, and several other languages.
Also, the rstanarm and brms packages run Stan under the hood but allow you to fit Bayesian models with the same formula interface used in the lm, glm and glmer functions. This allows you to fit many types of Bayesian models without having to learn to write code in the Stan language.