The recently deployed DeepTau algorithm for the discrimination of taus from light flavor quark or gluon induced jets, electrons, or muons is an ideal example for the exploitation of modern deep learning neural network techniques. With the current algorithm a suppression of misidentification rates by factors of two and more have been achieved for the same identification efficiency for taus compared to the MVA identification algorithms used for the LHC Run-1, leading to significant performance gains for many tau related analyses. The algorithm and its performance will be discussed.