Your question is State-of-the-Art ML Problem Solution. 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).
Explain how you would build a state-of-the-art solution to an unsolved, actively researched ML problem.
Asked in the ML domain stage for an ML Engineering Manager role. Treat the problem as open-ended: choose a suitable research problem, clarify its objective and constraints, and explain how you would take it from data collection through reliable production evaluation. Discuss what makes the problem unsolved, how you would manage scientific uncertainty, and how you would balance research novelty with engineering practicality.