Solution Overview

Deployment Scheme

For model protection in edge inference scenarios, there are logically the user management node, running node, and AI-GUARD (user host) node. They have the following functions:

• User management node: Provides central key services, controls master keys and pre-shared keys, and deploys encryption and decryption working keys.
• Running node: Executes inference jobs.
• AI-GUARD node: Provides the CLI encryption tool to encrypt data for local use.