In this chapter, we will look at ModelOps and its closest cousin—DevOps. We will explore how to build development pipelines for data science and make projects reliable, experiments reproducible, and deployments fast. To do this, we will familiarize ourselves with the general model training pipeline, and see how data science projects differ from software projects from the development infrastructure perspective. We will see what tools can help to version data, track experiments, automate testing, and manage Python environments. Using these tools, you will be able to create a complete ModelOps pipeline, which will automate the delivery of new model versions, while taking care of reproducibility and code quality.
In this chapter, we will cover the following topics:
- Understanding ModelOps
- Looking into DevOps
- Managing code versions and quality
- Storing data...
