Review
- Author: @woodword-0
- Repo: Azure/mlops-v2
- Title: Azure mlops-v2 Solution Accelerator
The Azure mlops-v2 solution accelerator is not only easy to use, but also a nice addition to any senior data scientist’s toolbox.
This repository allows a team of data scientist and data engineers to quickly deploy and manage a multitude of ML models following core principles of responsible AI.
One downside to this repository is the fact that it requires various roles for deployment such as data engineers, data scientists and MlOps, and, as such, may not be suitable for smaller teams.
Another win for the “minus column” is the fact that you need to register for an Azure account.
The repository itself is fairly user-friendly offering heaps of documentation and links to helpful videos. It navigates smoothly with lots of guidance on cloud application frameworks,
deployment and Mlops. Moreover, everything can be implemented directly into a single shell script file. However, be forewarned, this mass of helpful information is not only a strength but also the bane of this repository.
The links and tutorials add up and can lead the user into a deep rabbit hole. With that said, if one just sticks to the basic premise of the repository, which can be deciphered in the ‘sparse_checkout.sh’ file,
the repository can give a great scaffold upon which one can create MLOps pipelines.
repository_ownerandrepository_namemust reference a real GitHub repo link:- ie.
https://github.com/{repository_owner}/{repository_name}
- ie.
-
Please don’t modify any files except this one.
- To add pictures, screenshot them and paste them directly into this file. GitHub will automagically convert them to a unique local asset.