Return to Projects



TrustyAI is an open-source toolkit designed for responsible AI. It provides features such as fairness and data drift metrics, explainable AI algorithms, and various other tools available at a library level. In addition, TrustyAI offers a containerized service, Kubernetes/OpenShift deployment, and integrations with Open Data Hub, ModelMesh, and KServe.

 TrustyAI consists of several components:

  • TrustyAI core: the library containing the metrics, AI explainers, and other utilities
  • TrustyAI service: a microservice to provide runtime support for fairness metrics/explainability algorithms
  • TrustyAI operator: a Kubernetes operator responsible for the TrustyAI service lifecycle management
  • Python TrustyAI: a library allowing the usage of TrustyAI from Python
See the project