[BjoernKW]

Some currently spend a lot of their time on servicing and properly orchestrating Docker and related infrastructure (the actual product they're working on, mark you, isn't Docker-related ...).

On a much more positive note though a particularly smart person I know works on speech recognition software that uses linguistic features in order to assess communication skills and to some extent psychological properties (in an anonymous, privacy-compliant fashion) such as current predisposition for depression (which can manifest in features like pitch, voice melody and prosody).