Model maintenance

MLOps Deployment and Life Cycling

Nemanja Radojkovic

Senior Machine Learning Engineer

Previously...

Focus of this lesson: ML model failure

model deteriorated

improve the model

improve dataset

model centric

data centric

¹ https://datacentricai.org/

Label == Target variable value in the training set

Label quality == label closeness to the ground truth

poor labels

poor prediction

Manual labeling is complex, lengthy, error-prone

Use good labeling tools!

labeling tool

Example: Image labeling tool, for building image classifiers

human in the loop

HIL 2

expert input

Run ML build pipeline => New model
Check if new model is better than the old one
- YES => test and deploy
- NO => keep searching,

Immensely helpful: Metadata store (MLFlow Tracking, etc)

In any case: MLOps helps us maintain our model in the fastest, most efficient way

MLOps Deployment and Life Cycling