How to Store MLOps Pipeline Artifacts for Dozens of Pipelines

PMLE By islever Editorial Last updated

Tracking and running ML experiments. Developing end-to-end ML pipelines.

Short answer

C Store pipeline parameters in Vertex ML Metadata, source code in GitHub, and model binaries in Cloud Storage.

Question

You are building a MLOps platform to automate your company’s ML experiments and model retraining. You need to organize the artifacts for dozens of pipelines. How should you store the pipelines’ artifacts?
  1. Store parameters in Cloud SQL, and store the models’ source code and binaries in GitHub.
  2. Store parameters in Cloud SQL, store the models’ source code in GitHub, and store the models’ binaries in Cloud Storage.
  3. Store parameters in Vertex ML Metadata, store the models’ source code in GitHub, and store the models’ binaries in Cloud Storage. Correct Answer
  4. Store parameters in Vertex ML Metadata and store the models’ source code and binaries in GitHub.

Key Takeaways

The question tests matching each artifact type to the service designed for it; the trap is assuming any durable database such as Cloud SQL can replace Vertex ML Metadata for pipeline parameters and lineage.

When organizing MLOps artifacts across many pipelines, each artifact type belongs in its purpose-built store: parameters in Vertex ML Metadata, source code in GitHub, and model binaries in Cloud Storage, which is answer C.

Common mistake: Option B is the most plausible wrong answer: it correctly uses GitHub for code and Cloud Storage for binaries, but puts parameters in Cloud SQL, which stores values without the run-to-artifact lineage needed to organize dozens of pipelines.

Expert Analysis

The Attractive Trap: Option B

Option B looks like the complete, modern answer. It correctly puts source code in GitHub for version control and model binaries in Cloud Storage — and a comment notes that Google recommends "Git for code versioning and Cloud Storage with an integrated Model Registry for binaries and artifacts." That satisfies two thirds of the question, which is why it is the most seductive choice. It fails on exactly one point: parameters in Cloud SQL. Cloud SQL is a general-purpose relational database; it can hold key-value pairs, but it records no run-to-artifact relationships and no lineage across pipelines. The question's real requirement is organizing artifacts "for dozens of pipelines" — that is a metadata and lineage problem, not a CRUD problem.

Why Option C Is Correct

Vertex ML Metadata is purpose-built for this job: it stores parameters, executions, inputs, outputs and their relationships as a graph across all pipeline runs, so you can trace which settings produced which model. As one comment frames it, each artifact type belongs in the tool built for it — parameters in ML Metadata, source code in GitHub, model binaries in Cloud Storage, which scales cheaply and integrates with Vertex AI's Model Registry.

Ruling Out A and D

Option A misplaces binaries in GitHub, and Option D repeats that mistake while also pushing source code into ML Metadata, which tracks pipeline metadata rather than repositories. Comments warn that Git is inefficient for large model files. Only Option C places all three artifact types correctly.

Official Reference

Exam Strategy

When an option splits storage across three artifact types, judge each type against its purpose-built service rather than a generic one: parameters need lineage, code needs version control, binaries need object storage. First eliminate any option that commits large model files to Git, then check whether the parameter store records run-to-artifact relationships. Apply the weakest-link rule — one misplaced artifact invalidates an otherwise plausible option, which is exactly how B is defeated.

Frequently Asked Questions

Why isn't Cloud SQL a good place for pipeline parameters?

Cloud SQL can store values, but it has no built-in lineage linking a run to the datasets and models it produced. Vertex ML Metadata records parameters, executions, inputs and outputs as a graph, which is what organizing dozens of pipelines requires.

Can model binaries be committed to GitHub?

Git is designed for text source history, not large binary files; committing model artifacts bloats repositories and slows clones. Store binaries in Cloud Storage and register them with Vertex AI Model Registry instead.

What does Vertex ML Metadata actually track?

It tracks pipeline artifacts, executions, parameters and their relationships as lineage across runs, so you can trace which data and settings produced a given model.

More PMLE FAQ →

Community Votes

C
100%

100% of anonymous learners picked answer C. Votes are pick records left by other test-takers — they are not the verified answer.

Discussion 7 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 7

  1. fitri001 👍 3 Answered C
    Vertex ML Metadata: This service is specifically designed to store and track metadata for ML pipelines, including parameters. It provides a centralized location to manage and query pipeline execution details, making it ideal for dozens of pipelines. Cloud Storage: This is a scalable and cost-effective storage solution for model binaries. It integrates well with Vertex AI and other cloud services. GitHub: While not a Google Cloud service, it's a popular version control system well-suited for storing and managing your models' source code, particularly for collaboration among team members.
  2. pikachu007 👍 2 Answered C
    A. Cloud SQL and GitHub: Cloud SQL isn't designed for ML metadata management, potentially leading to challenges in tracking experiment details and lineage. B. Cloud SQL, GitHub, and Cloud Storage: While viable, this approach misses the benefits of Vertex ML Metadata for organized ML artifact management. D. Vertex ML Metadata and GitHub: Storing model binaries in GitHub can be inefficient for large files and might incur higher storage costs.
  3. OpenKnowledge 👍 2 Answered C
    For storing machine learning model source code and binaries, Google recommends using Git for code versioning and Cloud Storage with an integrated Model Registry for binaries and artifacts.
  4. shadz10 👍 2 Answered C
    https://cloud.google.com/architecture/architecture-for-mlops-using-tfx-kubeflow-pipelines-and-cloud-build
  5. taylz876 👍 1 Answered C
    Why C is right Each type of artifact goes in the tool built for it: • Parameters, in ML Metadata: it is the central store that records parameters, runs, datasets and lineage across all your pipelines as a graph, so you can trace which data and settings produced which model. That matters with dozens of pipelines. • Source code, in GitHub: this is version control, with history, branches and reviews. GitHub also connects to CI/CD triggers like Cloud Build. • Model binaries, in Cloud Storage: large files such as model_v3.pth are stored in Cloud Storage, which is cheap and scalable. Why the others are wrong • A and B: Cloud SQL is a relational database for transactional data. It isn't built for tracking ML parameters and lineage across pipelines. A also puts large model binaries in GitHub, which is for source code, not big binary files. • D: ML Metadata is right for the parameters, but GitHub isn't the right home for model binaries. Large binary files bloat a repository and are better kept in Cloud Storage.
Show 2 more comments
  1. fitri001 👍 1
    A. Cloud SQL for Parameters: While Cloud SQL is a relational database service, Vertex ML Metadata offers a dedicated solution for ML metadata management, including parameters, providing better integration and functionality within the MLOps context. D. Vertex ML Metadata for Source Code and Binaries: Vertex ML Metadata is primarily focused on ML pipeline metadata and experiment tracking. Cloud Storage is a more appropriate service for storing large binary files like model artifacts.
  2. pinimichele01 👍 1 Answered C
    shadz10

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Practice All PMLE Questions

Access 65 questions with complete answers and detailed explanations.

View Full PMLE Practice Test →

← Back to PMLE Study Guide