My job is mostly machine learning and statistics. I would love to use something like Haskell or even Java, but the NumPy/SciPy/Scikit-learn/Pandas etc. ecosystem is just so far ahead of everything else that it's not worth it.