This repository organizes some of the best practices in deploying ML projects using AWS cloud offerings. Each topic will have conceptual explanations of each example code.
The following topics are WIP and covered.
- How to do deployments using built-in AWS ML Algorithms.
- How to do deployments using AWS DL containers and package your own algorithms
- How to do deployments using custom containers (including extending AWS DL containers) and your own algorithms
- Resource usage profiling a deployment endpoint - a library of APIs that leverage on CloudWatch metrics
- Load testing a deployment endpoint showcasing how auto-scaling helps - using serverless artillery
- An AWS ML instance right sizing recommender for deployments - library that combines the above two
- Examples on how to profile deep learning architectures in TF, MXNet and PyTorch
- Examples of Deployment on the Edge
- Examples of using Sagemaker Model monitoring in deployment pipelines
- Examples of Computer Vision based deployment use cases
- Examples of NLP based deployment use cases
- Examples of Forecasting based deployment use cases
- Examples of Recommender systems based deployment use cases
- Examples of Reinforcement Learning based deployment use cases