Your question is Bias-Variance in Ensemble Tuning. 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 do you approach the bias-variance trade-off when tuning a complex ensemble model?