The Simulation Process and Data Management module (“SPDM”) which enables management of multiple candidate models as well as traceability of simulation data, crucial for compliance, certification, validation, and data management, and fundamental to ensuring traceability across the product lifecycle.

A test bed for performing human-in-the-loop design-of-validation experiments for complex models for identification of ‘model vulnerabilities’, i.e. regions and regimes where model prediction is failing.

Enables analysis of models under uncertain inputs, providing analytics on sensitivity, uncertainty and model robustness.

This can connect to distributed computing networks and HPC for running simulations at scale.

Particularly for AI models, this supports interpretability and explainability of black-box subsystems within a model. Provides local post-hoc explainability methods (SHAP, LIME, Contrastive Explanations and ProtoDash), as well as Global post-hoc explainability methods (ProfWeight, etc), as well as Causal Network Discovery to enable the discovering of driving process within complex networks of interacting models.

Lives alongside deployed models and monitors drift from area of trust, provides continual error rates and performance metrics to provide post-deployment assessment of models.