Kubeflow Pipelines for Monthly End-to-End Retraining

PMLE By islever Editorial Last updated

Developing end-to-end ML pipelines. Automating model retraining.

Short answer

A Configure Kubeflow Pipelines to schedule the end-to-end monthly retraining workflow from training through deploying the real-time delay-estimation model.

Question

You work for a public transportation company and need to build a model to estimate delay times for multiple transportation routes. Predictions are served directly
to users in an app in real time. Because different seasons and population increases impact the data relevance, you will retrain the model every month. You want to
follow Google-recommended best practices. How should you configure the end-to-end architecture of the predictive model?
  1. Configure Kubeflow Pipelines to schedule your multi-step workflow from training to deploying your model. Correct Answer
  2. Use a model trained and deployed on BigQuery ML, and trigger retraining with the scheduled query feature in BigQuery.
  3. Write a Cloud Functions script that launches a training and deploying job on AI Platform that is triggered by Cloud Scheduler.
  4. Use Cloud Composer to programmatically schedule a Dataflow job that executes the workflow from training to deploying your model.

Key Takeaways

The trap is answering "how do I retrain monthly" instead of "how do I configure the end-to-end architecture": only option A spans training and deployment in one orchestrated pipeline.

This PMLE question tests how to architect an end-to-end, monthly-retrained model for real-time transit delay predictions. The Google-recommended choice is option A, Kubeflow Pipelines (the basis of Vertex AI Pipelines), because it orchestrates the full training-to-deployment workflow.

Common mistake: Option D is the tempting distractor: Cloud Composer is a legitimate scheduler, but pairing it with Dataflow does not run model training or deployment, so it never delivers a true end-to-end ML workflow.

Expert Analysis

The decisive qualifier: "end-to-end architecture" The question does not ask how to retrain alone; it asks how to configure the whole path from training through deployment, with monthly retraining and real-time app serving. Only option A describes a single orchestrated multi-step workflow that spans training and deploying, which is exactly the end-to-end scope the wording demands.

Why option B fails the end-to-end test BigQuery ML plus a scheduled query can retrain a model on a calendar, but a scheduled query only re-executes CREATE MODEL. It does not export, deploy or serve the refreshed model to your app, so another tool is still required — the "end-to-end" promise is unmet. It is also a weak fit for serving predictions "directly to users in an app in real time" across multiple routes.

Why option C and D are orchestration fragments Option C is a hand-rolled glue job: a Cloud Functions script triggered by Cloud Scheduler, launching a separate AI Platform job. That is plumbing you maintain yourself, not a Google-recommended pipeline pattern. Option D pairs Cloud Composer with Dataflow, but Dataflow is a batch/stream data-processing engine, not a model training and deployment runner; Composer would still have to invoke a training service, so the workflow is not natively end-to-end.

Matching "Google-recommended best practices" Google's ML best-practice guidance points to pipeline orchestration with Vertex AI Pipelines, which is built on Kubeflow Pipelines — so option A is the pattern the guidance names. Community comments converge on A for this reason, with the main objection being that the option says "Kubeflow" rather than "Vertex AI"; they are the same orchestration model here. One dissenting comment argues D, but it concedes Composer schedules rather than runs training, which confirms A.

Correct answer: option A

Official Reference

These links open the official vendor documentation in a new tab.

Exam Strategy

On architecture questions, underline the scope word — "end-to-end", "fully managed", "least cost" — and reject any option that covers only one stage of the lifecycle. Here, options B, C and D each cover retraining or orchestration but leave a manual gap before deployment, so A survives. When two options both schedule work, check whether the scheduled component can actually execute training and serving, not merely trigger it.

Frequently Asked Questions

Why isn't BigQuery ML with a scheduled query enough for this scenario?

A scheduled query only retrains the model; it does not export, deploy or serve the updated model to the app. You would still need a separate serving step, so it is not an end-to-end architecture.

Can Cloud Composer replace Kubeflow Pipelines for monthly retraining?

Cloud Composer (Apache Airflow) orchestrates and schedules jobs, but it does not natively run model training and deployment steps. Dataflow handles data processing, not training, so option D still requires extra tooling.

Is Kubeflow Pipelines the same as Vertex AI Pipelines?

Vertex AI Pipelines is Google's managed implementation of the Kubeflow Pipelines SDK, so configuring Kubeflow-style pipelines matches Google's recommended ML workflow orchestration guidance.

More PMLE FAQ →

Community Votes

A
83%
D
17%

83% of anonymous learners picked answer A. Votes are pick records left by other test-takers — they are not the verified answer.

Discussion 26 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 26

  1. Paul_Dirac 👍 43
    Answer: A A. Kubeflow Pipelines can form an end-to-end architecture (https://www.kubeflow.org/docs/components/pipelines/overview/pipelines-overview/) and deploy models. B. BigQuery ML can't offer an end-to-end architecture because it must use another tool, like AI Platform, for serving models at the end of the process (https://cloud.google.com/bigquery-ml/docs/export-model-tutorial#online_deployment_and_serving). C. Cloud Scheduler can trigger the first step in a pipeline, but then some orchestrator is needed to continue the remaining steps. Besides, having Cloud Scheduler alone can't ensure failure handling during pipeline execution. D. A Dataflow job can't deploy models, it must use AI Platform at the end instead.
  2. gcp2021go 👍 11
    the answer is D. found similar explaination in this course. open for discussion. I found B could also work, but the question asked for end-to end, thus I choose D in stead of B https://www.coursera.org/lecture/ml-pipelines-google-cloud/what-is-cloud-composer-CuXTQ
  3. celia20200410 👍 6
    ANS: A https://medium.com/google-cloud/how-to-build-an-end-to-end-propensity-to-purchase-solution-using-bigquery-ml-and-kubeflow-pipelines-cd4161f734d9#75c7 To automate this model-building process, you will orchestrate the pipeline using Kubeflow Pipelines, ‘a platform for building and deploying portable, scalable machine learning (ML) workflows based on Docker containers.’
  4. lordcenzin 👍 4
    yes you can but it is not supposed to do that. DF is for data processing and transformation. you would loose all shenanigans kubeflow provide as native. Among the two answers, i think A is the most correct
  5. george_ognyanov 👍 3
    Answer seems to be A really. Here is a link from Google-recommended best practices. They are talking about Vertex AI Pipelines, which are essentially Kubeflow. https://cloud.google.com/architecture/ml-on-gcp-best-practices?hl=en#machine-learning-workflow-orchestration
Show 21 more comments
  1. mousseUwU 👍 3
    I guess it's A
  2. tavva_prudhvi 👍 2
    D is incorrect. Cloud Composer is a fully managed workflow orchestration service built on Apache Airflow. It is a recommended way by Google to schedule continuous training jobs. But it isn’t used to run the training jobs. AI Platform is used for training and deployment.
  3. gcper 👍 2
    A Kubeflow can handle all of those things, including deploying to a model endpoint for real-time serving.
  4. q4exam 👍 2
    Dataflow can deploy model .... this is how you do stream inference on stream
  5. lordcenzin 👍 2 Answered A
    A for me too. KF provides all the end2end tools to perform what is asked
  6. John_Pongthorn 👍 2 Answered A
    A : Yet the newer is Vertext-AI Pipeline built on Kubeflow
  7. ms_lemon 👍 2
    But D doesn't follow Google best practices
  8. caohieu04 👍 2 Answered A
    Community vote
  9. harithacML 👍 1 Answered A
    Req: retrain the model every month+ Google-recommended best practice+ end-to-end architecture A. Configure Kubeflow Pipelines to schedule your multi-step workflow from training to deploying your model. : Supports all above B. Use a model trained and deployed on BigQuery ML, and trigger retraining with the scheduled query feature in BigQuery : Why BigQuery ML when vertexAI/kubflow can handle end to end. BigQuery ML+ traigger only initiate the code run. C. Write a Cloud Functions script that launches a training and deploying job on AI Platform that is triggered by Cloud Scheduler. : Not recommended by google for end to end ML D. Use Cloud Composer to programmatically schedule a Dataflow job that executes the workflow from training to deploying your model. : Not recommended by google for end to end ML. what if model fails? matrix monitor?
  10. Fatiy 👍 1 Answered A
    A : In this case, it would be a good fit as you need to retrain your model every month, which can be automated with Kubeflow Pipelines. This makes it easier to manage the entire process, from training to deploying, in a streamlined and scalable manner.
  11. AmitK121981 👍 1 Answered C
    why its called kubeflow pipeline in OPtion A, not Vertex AI pipeline, coz KFP can be generic, on any K8s or GKE or so.
  12. EFIGO 👍 1 Answered A
    A is correct All the options get you to the required result, but only A follows the Google-recommended best practices
  13. Sum_Sum 👍 1 Answered A
    D - Dataflow job can't deploy models B,C are not - are not complete solutions leaving A to be the correct one
  14. q4exam 👍 1
    I think both A and D are correct because it is just different fashion of doing ML ...
  15. abhi0706 👍 1
    Answer is A: Kubeflow Pipelines can form an end-to-end architecture
  16. PhilipKoku 👍 1 Answered A
    A) Kubeflow Pipelines is the answer.
  17. mousseUwU 👍 1
    Please send a source link?
  18. GCP72 👍 1 Answered A
    Correct answer is "A"
  19. suranga4 👍 1
    Answer is A
  20. M25 👍 1 Answered A
    Went with A
  21. fragkris 👍 1 Answered A
    Chose A

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.

How to Minimize Development Time for Weekly TensorFlow Retraining in Vertex AI?

S-grade

How to Handle Categorical Features in BigQuery ML for Fast Deployment?

S-grade

How to preprocess BigQuery data for TensorFlow?

S-grade

How to Track Model and Prediction Lineage in Vertex AI?

S-grade

Practice All PMLE Questions

Access 65 questions with complete answers and detailed explanations.

View Full PMLE Practice Test →

← Back to PMLE Study Guide