, machine learning is a hypercompetitive game of ensemble engineering. The difference of a slight improvement in lap time or loss scores can be measured in the millions of dollars a team brings in when they do what it takes to be the best. Not only does every single component of the system need to be perfect, the way it is all brought together needs to be perfect too. The state of the art Gradient boosted models have historically been the most competitive models for tabular and time series prediction problems. These are ensemble methods because they combine the results of several base estimators to come up with a final answer that is better than any individual prediction alone. But the state of the art is beginning to change. Pre-trained models such as TabPFN for tabular data, and Chronos for time series are beginning to match or exceed gradient boosted models on certain benchmarks .…