A data science team is building a churn-prediction model on a tabular dataset with 40 mixed numeric and categorical features. Exploratory analysis shows strong nonlinear relationships and complex feature interactions between customer usage metrics. A baseline linear model underfits, achieving poor validation performance regardless of regularization tuning. The team wants a single algorithm that natively captures nonlinear interactions, handles mixed feature types well, and typically delivers high accuracy on structured data with modest tuning effort. Which SageMaker built-in algorithm should they choose?