Building a primary MLSecOPs platform
We will apply our MLSecOps patterns using Jenkins and MLFlow, which are the core of our MLSecOps. The book’s repository contains all the necessary files, and in this section, we will walk through how to build a basic MLSecOPs platform and use it for our Foodie AI solution.
The following diagram illustrates the basic architecture of our simple MLSecOps platform:
Figure 18.1 – MLSecOps architecture
This is a sample architecture to demonstrate the concepts and help understand the technical challenges. It combines Jenkins pipelines, supporting MLSecOps scripts, and MLFlow:
- Jenkins pipelines are called with parameters via the Jenkins API in invoker scripts, securely invoking them from the CLI, Jupyter notebooks, Git hooks, and
cron-like schedules. The pipelines use stages and steps to orchestrate these scripts and fail if specific compliance baselines are violated. - Automation scripts implement...
