0PricingLogin
MLOps Academy · Lesson

Configure with ConfigMaps and Secrets

Externalize model config and credentials.

Keep Config Out of the Image

Baking model paths or thresholds into the container forces a rebuild for every tweak. Kubernetes lets you inject config at runtime instead. 🔧

ConfigMaps Hold Plain Settings

A ConfigMap stores non-secret key-value pairs: a model name, a batch size, a log level. One config object, many Pods reading from it.

apiVersion: v1
kind: ConfigMap
metadata:
  name: model-config
data:
  MODEL_NAME: "ranker-v3"
  BATCH_SIZE: "32"

All lessons in this course

  1. Pods, Deployments, and Services for Models
  2. Request CPU, Memory, and GPU
  3. Configure with ConfigMaps and Secrets
  4. Run Training as a Kubernetes Job
← Back to MLOps Academy