DevOps Practices in Machine Learning

What is Continuous Integration in Machine Learning?

• Continuous Integration in Machine Learning is the practice of frequently integrating code changes and automating testing and validation of models.

• Continuous Integration is the process of collecting data for training models.

• Continuous Integration is only about deploying models to production.

• Continuous Integration in Machine Learning refers to the manual testing of models.

Name two popular tools used for CI in ML projects.

• Travis CI

• CircleCI

• Bamboo

• Jenkins, GitLab CI

What are the key considerations for model deployment strategies?

• Cost-effectiveness

• Data storage solutions

• Scalability, reliability, monitoring, security, ease of integration.

• User interface design

Describe the difference between blue-green deployment and canary deployment.

• Blue-green deployment is only for testing, while canary deployment is used for production releases.

• Blue-green deployment is a single environment approach, while canary deployment uses multiple environments.

• Blue-green deployment switches between two complete environments, while canary deployment gradually rolls out changes to a small user group.

• Canary deployment switches between two complete environments, while blue-green deployment gradually rolls out changes.

Why is monitoring ML models important after deployment?

• Monitoring ML models is primarily for compliance with regulations.

• Once deployed, ML models do not require any further evaluation.

• Monitoring ML models is only necessary during the initial training phase.

• Monitoring ML models is important to ensure ongoing performance, detect issues, and maintain accuracy.

List three metrics that can be used to monitor ML model performance.

• Mean Squared Error

• F1 Score

• Accuracy, Precision, Recall

• AUC-ROC

How does version control benefit machine learning projects?

• Version control is only useful for software development, not machine learning.

• It helps in data cleaning and preprocessing only.

• Version control benefits machine learning projects by enabling collaboration, tracking changes, ensuring reproducibility, and managing experiments effectively.

• It eliminates the need for documentation in projects.

What are some common version control systems used in ML?

• TFS

• Git, DVC, MLflow, and Pachyderm

• Mercurial

• Subversion

Explain the role of automated testing in ML workflows.

• Automated testing focuses solely on user interface design.

• Automated testing is only used for software development.

• Automated testing ensures reliability, validates data integrity, checks model accuracy, and maintains performance consistency in ML workflows.

• Automated testing eliminates the need for human oversight in ML.

What is Infrastructure as Code and how is it applied in ML?

• IaC is used to manually configure servers for ML applications.

• Infrastructure as Code is a programming language for data analysis in ML.

• Infrastructure as Code (IaC) is a practice for managing infrastructure through code, applied in ML to automate environment setup for training and deployment.

• Infrastructure as Code is a method for writing machine learning algorithms directly.