The article introduces "quantile-forest," a Python package for implementing quantile regression forests (QRF), a non-parametric, tree-based ensemble method for estimating conditional quantiles. QRF is an extension of the random forests algorithm, which is widely used for machine learning tasks. Unlike random forests, QRF provides probabilistic predictions by outputting the weighted empirical distribution of training labels, making it useful for understanding uncertainty in regression problems. The package, optimized using Cython for speed, allows users to estimate arbitrary quantiles at prediction time without retraining. It includes features such as out-of-bag estimation, quantile rank calculation, and proximity counts, enhancing its applicability in various research and business settings. The package is designed to be compatible with scikit-learn and has been cited in several scholarly works and used in production at Zillow Group. The article highlights the need for a comprehensive Python implementation of QRF, as existing R implementations are more widely used. The Python package addresses this gap by providing a fast, feature-rich implementation that scales well to large datasets and offers utilities for model evaluation and data analysis tasks. Examples are provided to demonstrate how to train and predict using the package, estimate quantile ranks, and compute proximities. The package is intended to empower researchers and practitioners to gain deeper insights into complex data by accurately estimating conditional quantiles and understanding the underlying data distribution.The article introduces "quantile-forest," a Python package for implementing quantile regression forests (QRF), a non-parametric, tree-based ensemble method for estimating conditional quantiles. QRF is an extension of the random forests algorithm, which is widely used for machine learning tasks. Unlike random forests, QRF provides probabilistic predictions by outputting the weighted empirical distribution of training labels, making it useful for understanding uncertainty in regression problems. The package, optimized using Cython for speed, allows users to estimate arbitrary quantiles at prediction time without retraining. It includes features such as out-of-bag estimation, quantile rank calculation, and proximity counts, enhancing its applicability in various research and business settings. The package is designed to be compatible with scikit-learn and has been cited in several scholarly works and used in production at Zillow Group. The article highlights the need for a comprehensive Python implementation of QRF, as existing R implementations are more widely used. The Python package addresses this gap by providing a fast, feature-rich implementation that scales well to large datasets and offers utilities for model evaluation and data analysis tasks. Examples are provided to demonstrate how to train and predict using the package, estimate quantile ranks, and compute proximities. The package is intended to empower researchers and practitioners to gain deeper insights into complex data by accurately estimating conditional quantiles and understanding the underlying data distribution.