❗Disclaimer: All opinions are my own.
Overview
A full open-source solution for protecting AI/ML code and data owned by party A (the provider) from being viewed in plaintext form by party B (the user). Party B can use the ML application on its own infrastructure, but cannot view the intellectual property.
Solution
Referencing my materials below as I’m the main author.
Blog post: https://aws-samples.github.io/aws-graviton-run-confidential-ml-workloads-using-nitro-enclaves/
Solution code: https://github.com/aws-samples/aws-graviton-run-confidential-ml-workloads-using-nitro-enclaves