Your question is Debugging Inference Latency. 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).
Given a large, complex codebase, what is your strategy for debugging an inference latency issue in an edge ML system like Syntiant?