Your question is Optimize an On-Device Neural Network. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You are building a neural network that runs directly on mobile devices for a user-facing product. The current model quality is acceptable, but inference cost, memory use, and battery impact make the experience hard to ship broadly.
Explain how you would approach optimizing a neural network for efficiency on mobile devices.