Your question is Handling Traffic Light Malfunctions. 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).
How would you handle an expected traffic light malfunction in Zoox's autonomy stack?
Treat this as an end-to-end ML system design problem for Zoox vehicles operating at intersections. Explain how you would detect abnormal or unreliable traffic-light behavior, combine perception with map and vehicle context, select a safe behavior under uncertainty, and validate the system offline and on-road. Address online inference, fallback behavior, monitoring, retraining, and failure containment without assuming that a malfunction can be directly observed in advance.