How to audit custom ML models using Amazon SageMaker

AWS Machine Learning - Model Governance and Ethics

An ML research team develops custom ML models. The model artifacts are shared with other teams for integration into products and services. The ML team retains the model training code and data. The ML team wants to build a mechanism that the ML team can use to audit models. Which solution should the ML team use when publishing the custom ML models?

Create documents with the relevant information. Store the documents in Amazon S3.
Use AWS AI Service Cards for transparency and understanding models.
Create Amazon SageMaker Model Cards with intended uses and training and inference details. Source Reference Answer
Create model training scripts. Commit the model training scripts to a Git repository.

Community Votes

100%

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

Community Insight

The question tests knowledge of AWS-specific tools for ML governance; the trap is choosing generic documentation methods (S3/Git) instead of the specialized SageMaker Model Cards feature designed for this purpose.

Amazon SageMaker Model Cards provide a standardized mechanism for documenting model details, intended uses, and training data to facilitate auditing. The community consensus confirms this is the correct solution for transparency and compliance in ML workflows.

Candidates often choose Option A (Amazon S3) because it is a common storage solution, but they fail to recognize that simple file storage does not provide the structured, standardized metadata required for effective model auditing and transparency.

Community Discussion (5 comments)

kopper2019 👍 1

C. Create Amazon SageMaker Model Cards with intended uses and training and inference details.

Jessiii 👍 1 Selected: C

Amazon SageMaker Model Cards are designed to provide transparency and detailed information about a model's intended uses, training process, data, and performance. They are specifically designed for auditing models and making them more interpretable, which is exactly what the ML team needs. Model cards help document key details about the model, including how it was trained, its intended use cases, and any potential risks associated with its use. This solution supports effective auditing and provides a comprehensive, structured format for tracking important model information.

Moon 👍 2 Selected: C

Amazon SageMaker Model Cards: These provide a standardized way to document important information about ML models, including: Model purpose and intended use cases Training data and methodology Evaluation metrics and results Ethical considerations and limitations Bias analysis Version history This comprehensive documentation facilitates auditing by providing a clear record of how the model was developed, evaluated, and intended to be used. It also promotes transparency and accountability. B. Use AWS AI Service Cards for transparency and understanding models: AWS AI Service Cards are designed for pre-built AI services provided by AWS, not for custom ML models developed by a team. They are not applicable for this use case.

may2021_r 👍 1 Selected: C

The correct answer is C. SageMaker Model Cards are designed specifically for documenting model details and usage.

aws_Tamilan 👍 1 Selected: C

The correct answer is: C. Create Amazon SageMaker Model Cards with intended uses and training and inference details. Explanation: Amazon SageMaker Model Cards provide a standardized and centralized way to document key details about a machine learning model. This includes intended use, training and inference details, performance metrics, and ethical considerations. These cards enable the ML team to maintain transparency, track audit details, and share relevant information with other teams when publishing models.

Comments & Corrections

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

Expert Analysis

Why the Answer Is Correct

Amazon SageMaker Model Cards are specifically designed to document key aspects of a machine learning model, including its intended use cases, training methodology, evaluation metrics, and ethical considerations. By creating these cards, the ML team establishes a transparent record that facilitates auditing, ensuring that other teams understand how the model was built and its limitations.

Why the Other Options Are Wrong

Option A (S3 documents) lacks the structured format and specific fields required for comprehensive model governance. Option B refers to 'AI Service Cards,' which is not a standard AWS product name for custom model documentation. Option D (Git scripts) tracks code versioning but does not document the model's performance, data sources, or ethical implications needed for an audit.

Community Comment Notes

Comments [1], [3], and [5] strongly support Option C, highlighting that SageMaker Model Cards provide the necessary standardized documentation for transparency. Comment [4] reinforces that these cards are the specific tool for documenting model details and usage, aligning with the requirement to build an auditing mechanism.

Official Reference

https://docs.aws.amazon.com/sagemaker/latest/dg/model-cards.html

Exam Strategy

When encountering questions about ML governance, ethics, or auditing on AWS exams, immediately look for 'Model Cards' as the primary answer. Differentiate between code versioning (Git) and model metadata documentation (Model Cards), as the latter is required for transparency and auditability.

Related Analysis

Practice All AIF-C01 Questions

Access 100 questions with complete answers and detailed explanations.

View Full AIF-C01 Practice Test →

← Back to AIF-C01 Study Guide