Most of researchers these days prototype using Python & R. However, when you put ML systems into productions, accuracy is not the only metric. Teams care about scalability and the speed of their system 1

What do you use when you deploy ML in production ?. Which technology make it easier for you to build faster and more reliable infrastructure.

Most of answers to this question on [reddit](https://www.reddit.com/r/MachineLearning/comments/7zizn3/d_python_scala_rust_or_go_what_do_you_use_when/). I am curious to see more diverse answers and why teams tend to use specific technology for ML ?