Model governance

MLOps Deployment and Life Cycling

Nemanja Radojkovic

Senior Machine Learning Engineer

ai headlines

MLOps Deployment and Life Cycling

credit risk

MLOps Deployment and Life Cycling

underestimation

MLOps Deployment and Life Cycling

bankrupt

MLOps Deployment and Life Cycling

cost of decision 1

MLOps Deployment and Life Cycling

cost of decision 2

MLOps Deployment and Life Cycling

Governance

AI/ML model governance is the overall process for how an organization controls access, implements policy, and tracks activity for models and their results.

Effective model governance is the bedrock for minimizing risk to both an organization's bottom line and to its brand.

~ DataRobot.com

MLOps Deployment and Life Cycling

governance phases

MLOps Deployment and Life Cycling

design

MLOps Deployment and Life Cycling

dev governance

MLOps Deployment and Life Cycling

ops governance

MLOps Deployment and Life Cycling

proportional to risk

MLOps Deployment and Life Cycling

use case based risk

MLOps Deployment and Life Cycling

Risk categories

MLOps Deployment and Life Cycling

Summary

  • Governance means extra steps, but alternative is anarchy.
  • "Launching many models fast" is not our goal, but generating business value.
  • Reckless ML --> more damage than benefit.
  • More models, more obvious the benefit of governance becomes.
MLOps Deployment and Life Cycling

Let's practice!

MLOps Deployment and Life Cycling

Preparing Video For Download...