Your question is Improving Model Quality with 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).
You're comparing a few supervised learning models and the current version is not meeting quality expectations on validation data. You need to improve generalization without changing the core problem setup.
How would you adjust hyperparameters to improve model quality?