Genetic engineers sound a lot like machine learning engineers— tweaking huge complicated models which can only be understood through observation/experimentation. Often times the training is stochastic and not directly repeatable too.