nerdexam
Google

PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #221

You have trained a model by using data that was preprocessed in a batch Dataflow pipeline. Your use case requires real-time inference. You want to ensure that the data preprocessing logic is applied…

The correct answer is B. Refactor the transformation code in the batch data pipeline so that it can be used outside of the. To ensure consistent data preprocessing for models trained with batch Dataflow and served for real-time inference, the transformation logic must be reusable across both environments.

Submitted by satoshi_tk· Apr 18, 2026ML pipeline operationalization

Question

You have trained a model by using data that was preprocessed in a batch Dataflow pipeline. Your use case requires real-time inference. You want to ensure that the data preprocessing logic is applied consistently between training and serving. What should you do?

Options

  • APerform data validation to ensure that the input data to the pipeline is the same format as the
  • BRefactor the transformation code in the batch data pipeline so that it can be used outside of the
  • CRefactor the transformation code in the batch data pipeline so that it can be used outside of the
  • DBatch the real-time requests by using a time window and then use the Dataflow pipeline to

How the community answered

(35 responses)
  • A
    9% (3)
  • B
    86% (30)
  • C
    3% (1)
  • D
    3% (1)

Why each option

To ensure consistent data preprocessing for models trained with batch Dataflow and served for real-time inference, the transformation logic must be reusable across both environments.

APerform data validation to ensure that the input data to the pipeline is the same format as the

Data validation ensures input quality but does not guarantee that the same preprocessing transformations are applied consistently between training and serving.

BRefactor the transformation code in the batch data pipeline so that it can be used outside of theCorrect

The core transformation code from the batch Dataflow pipeline needs to be extracted and made available to the real-time serving environment. This ensures that the exact same logic is applied to data before training and before real-time prediction, preventing training-serving skew. By refactoring the code, it can be integrated into the serving application without relying on the Dataflow pipeline itself for real-time processing.

CRefactor the transformation code in the batch data pipeline so that it can be used outside of the

This choice is identical to B, implying the same correct technical reason for refactoring the code.

DBatch the real-time requests by using a time window and then use the Dataflow pipeline to

Batching real-time requests and using Dataflow for processing would introduce significant latency, making it unsuitable for real-time inference requirements.

Concept tested: Training-serving skew prevention, data preprocessing consistency

Source: https://cloud.google.com/architecture/overview-ml-training-serving-skew

Topics

#Training-serving skew#Feature consistency#Real-time inference#Data preprocessing

Community Discussion

No community discussion yet for this question.

Sign up to join the discussion
Full PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice