LightGBM MLOps with AWS SageMaker
In Chapter 8, Machine Learning Pipelines and MLOps with LightGBM, we built an end-to-end ML pipeline using scikit-learn. We also looked at encapsulating the pipeline within a REST API and deployed our API to the cloud.
This chapter will look at developing and deploying a pipeline using Amazon SageMaker. SageMaker is a complete set of production services for developing, hosting, monitoring, and maintaining ML solutions provided by Amazon Web Services (AWS).
We’ll expand our capabilities with ML pipelines by looking at advanced topics such as detecting bias in a trained model and automating deployment to fully scalable, serverless web endpoints.
The following main topics will be covered in this chapter:
- An introduction to AWS and SageMaker
- Model explainability and bias
- Building an end-to-end pipeline with SageMaker
