3.1 Prototype

Research (If the DVC + MLFlow + Anaconda stack works for you, that's great. Metaflow provides similar features. Cloud integration is really important at Netflix's scale

https://news.ycombinator.com/item?id=21702831)

• Hardware: how to connect multiple servers

• Utility functions: the pre-processing/post-processing pipeline(DAG), the training pipeline, the inference pipeline, visualization pipeline,

• Model register: a platform to track model and compare model

• Data: Data management system/ data warehouse

• Multiple containers/env maintaining

What elements should be covered in an ML platform?

• Data management

• Experiment management

• Model management

• Deployment for inference

• Reproducibility

• Testing and monitoring

All through a consistent interface.