pht-ml trains deep learning models to classify planetary transits from TESS light curves using preprocessed PDCSAP fluxes from .fits files. The pipeline applies binning, normalization, and optional data augmentations, and it can incorporate external synthetic light curves alongside label files. Code in the data module handles loading of light curves and label tables when provided. Experiments and hyperparameters are configured via command-line arguments and run through a main training script.