Can InferTrust™ Work With Any AI/ML Framework?

Yes. InferTrust™ is framework-agnostic. It operates on the output of your model, not the internals. It works with TensorFlow, PyTorch, ONNX Runtime, scikit-lear

Framework-Agnostic by Design

InferTrust™ does not need to understand your model's architecture, training methodology, or internal representations. It operates at the inference boundary, the point where the model produces its output. This means it works with any framework that produces a deterministic output from a given input.

Supported Frameworks and Runtimes

• TensorFlow / TensorFlow Lite for cloud and edge deployments
• PyTorch / TorchScript for research and production workloads
• ONNX Runtime for cross-platform inference
• scikit-learn for traditional ML models
• Custom inference engines built in C++, Rust, or other languages
• Third-party model APIs including hosted LLM services, computer vision APIs, and NLP endpoints

How It Works Across Frameworks

The InferTrust™ SDK wraps the inference call regardless of the underlying framework. Your model produces its output, the SDK captures it at the boundary, constructs the seven-field decision record, signs it with the hardware key, and returns the result to your application. The framework is irrelevant to the signing process because the SDK operates on the output, not the computation.