I work at ModelFront, the machine translation quality prediction API.

We build machine learning that predicts if a translation is good or bad.

The most interesting application is predicting which machine translations do not require human editing, to translate content radically more efficiently, at human quality.