At this point in time, CFD and the types of tools used in engine design do not lend themselves very well to applications using GA. CFD and other computational tools are very slow (on the order of hours to days) to run and still require a lot of human oversight to produce decent results (so much for physics-based tools, eh?). In addition, it's not quite as simple as a CFD code. It will also have to include the thermodynamic cycle analysis, structural analysis, and noise analysis tools (and dozens of other tools in order to properly assess a possible design). In a very large design space, this would make a zero-order random search method very inefficient. GA are best when the evaluation of the fitness function is computationally inexpensive. That's not to say that there are no applications of GA in aerospace design, but right now they are much more limited in scope.

Engine design is incredibly complex. The big engine manufacturers spend several BILLION dollars designing the "core" of the engine alone.