Pipeline components and DAGs in MLOps - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is a pipeline in MLOps?

A pipeline is a series of connected steps that process data and train models automatically, like an assembly line in a factory.

beginner

What does DAG stand for and why is it important in pipelines?

DAG stands for Directed Acyclic Graph. It shows the order of steps in a pipeline without loops, ensuring tasks run in the right sequence.

beginner

Name three common components of an MLOps pipeline.

Data ingestion, data processing, and model training are three common pipeline components.

intermediate

How does a DAG help prevent errors in pipeline execution?

By defining a clear order without cycles, a DAG prevents tasks from running before their dependencies, avoiding confusion and errors.

intermediate

What happens if a pipeline step fails in a DAG-based system?

The pipeline stops or retries the failed step, preventing later steps from running with bad data or incomplete results.

What does a pipeline component NOT typically include?

A User interface design

B Data cleaning

C Model deployment

D Model training

User interface design is not part of pipeline components; pipelines focus on data and model steps.

Why must a DAG be acyclic?

A To allow tasks to run in parallel

B To speed up the pipeline

C To avoid infinite loops in task execution

D To reduce storage needs

A DAG must have no cycles to prevent infinite loops and ensure tasks finish.

Which component typically comes first in an MLOps pipeline?

A Model evaluation

B Feature engineering

C Model deployment

D Data ingestion

Data ingestion is the first step to bring data into the pipeline.

What is the main role of a DAG in pipeline management?

A To schedule tasks in order

B To store data

C To visualize model accuracy

D To monitor hardware usage

A DAG schedules tasks so they run in the correct order.

If a pipeline step depends on another, what does the DAG ensure?

A Both steps run simultaneously

B The dependency runs before the dependent step

C The dependent step runs first

D The steps run randomly

The DAG ensures dependencies run before dependent steps.

Explain what a pipeline is and describe the role of DAGs in managing pipeline steps.

Think of a pipeline as a recipe and DAG as the step-by-step instructions.

List common components of an MLOps pipeline and explain why the order of these components matters.

Consider what happens if you train a model before cleaning data.