Incorporating Continuous Integration and Continuous Delivery (CI/CD) in AI project pipelines

less than 1 minute read

Published:

CI/CD in AI projects

Either you are working with traditional machine learning (ML) models, or heavy deep learning (DL) models or LLM-based API requests, at the end of the day, you still need to deploy the models to (hopefully) a lot of users. One of the key elements in the deployment process, is CI/CD. This blog post is to introduce the idea of incorporating CI/CD Continuous Integration/Countinous Deployment in a AI-based software projects. I will first explain what CI/CD is, why we need it, what some key elements in CI/CD are, and last, show you an example of using CI/CD in a simple ML project where I will build/test/deploy a diabete diagnosis model on a really simple web-app with a small virtual machine on AWS.

(In progress)

Introduction

1. What is CI/CD?

2. Why do we need CI/CD?

3. Now how can we incorporate it in a project?

4. Future trend (my opinions) on CI/CD