CI/CD for ML pipelines in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding CI/CD stages in ML pipelines

Which stage in a CI/CD pipeline for ML is responsible for automatically retraining the model when new data arrives?

A Model deployment stage

B Continuous training stage

C Continuous integration stage

D Monitoring stage

💡 Hint

Think about which stage updates the model with fresh data.

The continuous training stage automatically retrains the ML model when new data is available, ensuring the model stays accurate over time.

💻 Command Output

intermediate

Output of a pipeline status command

What is the output of the command mlflow pipelines status --pipeline-name churn-prediction if the pipeline is currently running?

ML Python

mlflow pipelines status --pipeline-name churn-prediction

A Pipeline 'churn-prediction' status: RUNNING

B Error: Pipeline not found

C Pipeline 'churn-prediction' status: FAILED

D No pipelines currently running

💡 Hint

Look for the status message indicating the pipeline is active.

The command shows the current status of the specified pipeline. If it is running, the output will confirm that.

❓ Configuration

advanced

Correct YAML snippet for triggering retraining

Which YAML snippet correctly configures a GitHub Actions workflow to trigger model retraining on every push to the main branch?

ML Python

name: ML Retrain
on:
  push:
    branches:
      - main
jobs:
  retrain:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - name: Run retrain script
        run: python retrain.py

A

on:
  schedule:
    - cron: '0 0 * * *'

B

on:
  pull_request:
    branches:
      - main

C

on:
  push:
    branches:
      - main

D

on:
  push:
    tags:
      - main

💡 Hint

Look for the event that triggers on code push to the main branch.

Option C correctly triggers the workflow on every push to the main branch. Other options trigger on pull requests, tags, or scheduled times.

❓ Troubleshoot

advanced

Diagnosing failed model deployment

You have a CI/CD pipeline that builds and deploys an ML model. The build succeeds but deployment fails with the error: Permission denied: unable to push to model registry. What is the most likely cause?

A The model artifact is missing from the build

B The deployment script has a syntax error

C The model registry service is down

D The CI/CD pipeline lacks permissions to write to the model registry

💡 Hint

Consider access rights needed for deployment.

The error indicates the pipeline cannot push the model due to missing permissions. Fixing access rights resolves this.

🔀 Workflow

expert

Order of steps in a CI/CD ML pipeline

Arrange the following steps in the correct order for a typical CI/CD pipeline for ML models:

1. Deploy model to production
2. Run automated tests on model code
3. Train model with new data
4. Build model artifact

A 1,2,3,4

B 2,1,3,4

C 1,3,2,4

D 3,2,1,4

💡 Hint

Think about testing before training, then building, then deploying.

The correct order is to first run tests on code, then train the model, build the artifact, and finally deploy it.