Is there still value in learning/doing research in efficient interpreter implementation? Sooner or later you will need bytecode/JIT/AOT.