The Hyper-Parameter Heist

You are a member of a team of AI developers who are competing to create the best performing system using hyper-parameter optimization. Your team must navigate the complex world of machine learning algorithms, hyper-parameters and training procedures to build a model that outperforms your rivals. But beware, too much optimization of hyper-parameters on training data could lead to overfitting which will ultimately hurt your model's performance on the test data. Can you create an unbeatable system without cheating?