We often take this approach at my company. The heavy lifting of feature extraction from raw data (wearables in our case) are done by python/numpy models. The po

We often take this approach at my company. The heavy lifting of feature extraction from raw data (wearables in our case) are done by python/numpy models. The population level stuff is then often handled in R by data scientists with more of a maths/stats background than an engineering one.