Estimation and inference of a directed acyclic graph with unspecified interventions.

Modular and extensible Python framework for applying synthetic inference and controlled perturbations to AI model inputs, labels, and hyperparameters. Evaluate robustness, sensitivity, and stability of algorithms under realistic variations and adverse scenarios.