Comment by Myrmornis
9 hours ago
I'm curious about the involvement of tech companies here. Obviously approximating posterior distributions of explicit statistical models via simulation techniques is common in academic scientific literature but I'd like to hear about examples of it being done in "production" settings, i.e. not just as a one-off analysis. I have for a long time had a vague belief that in production settings people usually opt for heuristics / point estimates etc but I haven't had much involvement with this sort of thing for a while.
Pyro was created by Uber AI Labs. Actually, by Geometric Intelligence, which was eventually acquired by Uber. Geometric Intelligence was founded by Gary Marcus, Zoubin Ghahramani and others. They also had Noah Goodman onboard.
AFAIK, Pyro was used in production to make predictions of demand with careful consideration of uncertainty. I was contacted by one of their recruiters when I was doing work in this area, and this was the application they showcased.
Meta is also doing a lot of related work on time series forecasting using Prophet, which employs Stan under the hood. In both cases, Bayesian methods are important to make inference robust, it's not just an academic exercise.