R is good for one-off analysis problems, but is really bad for large distributed production systems. My team moved a large internal analysis application from R to Python because Python works well for both statisticians and software engineers.