The package offers fast and accurate convolution-type smoothed quantile regression. It is implemented using Barzilai-Borwein gradient descent with a Huber regression warm start. It constructs confidence intervals for regression coefficients using multiplier bootstrap.