How MLOps helps across all stages of ML project. By Alexey Grigorev.
It’s a common misconception that MLOps is solely about the tools we use for deploying models and preparing the infrastructure for it. Partly it is, but it’s not the whole story – there’s much more. In this post, I’ll break down a machine learning project into several stages and explain how MLOps helps at each of them.
Source @towardsdatascience.com: https://towardsdatascience.com/mlops-in-10-minutes-165c746a9b8e
MLOps is a new topic and there’s no consensus on what it is or what it is not. In this post, author will share his personal take on it. You don’t have to agree with it, but he hopes it’ll still be useful.
The article pays attention to:
- What is MLOps
- Train stage
- Experiment tracking
- Training pipelines
- Operate stage
- Deployment
- Model monitoring
- People, processes and best practices
In this article, we only scratched the surface. We discussed what MLOps is and looked at the helicopter view of the process. We broke down the process into 3 stages: design, train and operate. Good read!
[Read More]