Principles Of MLOps

Created: 2023-02-13 10:54
#note

Create an optimal end-to-end workflow that integrates different teams, modules, tools, and artifacts to continuously improve and deliver machine learning solutions.

Principles:

• modular;
• reproducible,
• integrated;
• continuous;
• scalable;
• responsive;
• automated;
• observable;
• transparent;
• incremental;
• traceable;
• collaborative;
• managed;
• secure;
• organized.

Requirements are fundamental for the success of a ML project:

• Non-ML requirements:
  • user experience;
  • solution functions;
  • deployment;
  • scale;
  • security;
  • serviceability;
• ML requirements:
  • function;
  • performance goals (effectiveness);
  • operational goals (efficiency);
  • cost.

References

1. LinkedIn

Tags

#mlops #course #ml