XGBoost (Extreme Gradient Boosting) is a high-performance, scalable implementation of gradient-boosted decision trees. It builds an ensemble of trees sequentially, where each new tree corrects the errors of the previous ones by focusing on samples that were harder to predict. It uses gradient descent to minimize a loss function and includes regularization terms to prevent overfitting.

XGBoost is known for its speed and accuracy, thanks to optimizations like parallelized tree construction, cache-aware data structures, and support for sparse data. It also handles missing values automatically and supports early stopping, cross-validation, and custom loss functions.

Widely used in structured data tasks (e.g., tabular data in finance, healthcare, or recommendation systems), XGBoost consistently performs well in machine learning competitions and production systems. It supports both classification and regression and has APIs in Python, R, Java, and other languages.

While powerful, XGBoost can be less interpretable than simpler models and may require careful hyperparameter tuning to perform optimally.