MLA-C01 AWS Certified Machine Learning Engineer - Associate Questions and Answers

Dynamic data masking allows you to control how sensitive data is presented to users at query time, without modifying or storing transformed versions of the source data. Amazon Redshift supports dynamic data masking, which can be implemented with minimal effort. This solution ensures that the data scientist can access the required information while sensitive data remains protected, meeting the requirements efficiently and with the least implementation effort.

To generate images, the application must invoke a foundation model hosted on Amazon Bedrock. Invoking the Amazon Titan Image Generator requires the bedrock:InvokeModel permission. Read-only permissions such as bedrock:Get* do not allow inference execution and are insufficient.

The application also interacts with Amazon Simple Queue Service. To process messages, the application must be able to receive messages from the queue and delete them after successful processing. These actions require sqs:ReceiveMessage and sqs:DeleteMessage.

sqs:GetQueueAttributes alone does not allow message consumption. SageMaker permissions are unrelated because Bedrock is a separate service.

Therefore, permissions to invoke the model and receive/delete SQS messages are required.

The issue is caused by oversized Parquet files that cannot be efficiently read into memory during training. The most effective and scalable solution is to repartition the dataset into smaller Parquet files.

AWS best practices for large-scale ML training recommend optimizing data layout, not simply increasing memory. By using Apache Spark on Amazon EMR, the ML engineer can repartition the Parquet files into smaller chunks that can be streamed and processed efficiently by SageMaker training jobs.

Attaching EBS volumes (Option A) increases storage capacity but does not solve in-memory constraints. Changing to memory-optimized instances (Option C) increases cost and does not address long-term scalability. SMDDP (Option D) distributes gradients and computation, not dataset file sizes.

Therefore, repartitioning the Parquet files is the correct solution.

Amazon EMR clusters consist of primary, core, and task nodes, each with a distinct role. The primary node manages the cluster, core nodes store data and run tasks, and task nodes only run tasks without storing data. AWS documentation recommends using task nodes for scaling compute capacity when workloads are compute-intensive, such as data ingestion and transformation pipelines.

To reduce processing time cost-effectively, AWS strongly advises using Spot Instances for task nodes. Spot Instances provide the same compute capacity as On-Demand Instances but at a significantly reduced cost, often up to 90% lower. Because task nodes do not store HDFS data, they can be safely interrupted without risking data loss.

Increasing the number of primary nodes is not supported by EMR and would not improve performance. Increasing core nodes affects both storage and compute and is more expensive, especially when using On-Demand Instances. Option D is therefore the least cost-effective.

AWS EMR best practices explicitly state that scaling out with Spot task nodes is the preferred way to improve performance for transient, parallel workloads such as ETL, ingestion, and feature preparation.

Therefore, Option C is the most cost-effective and AWS-recommended solution.

Scenario: The dataset contains duplicate records that need to be detected with minimal code development.

Why FindMatches in AWS Glue?

Purpose-Built for Deduplication: The FindMatches transform in AWS Glue is specifically designed to identify duplicate records in structured or semi-structured datasets.

Machine Learning-Based: It uses ML to identify duplicates based on configurable thresholds and provides flexibility for tuning accuracy.

Low Code Overhead: Minimal development effort is required as Glue provides an interactive console for configuring and running FindMatches transforms.

Steps to Implement:

Prepare the Data: Upload the unstructured dataset to an S3 bucket and define a schema if needed.

Create a Glue Job:

Use the AWS Glue Studio to create a job and select the FindMatches transform.

Specify key attributes for deduplication.

Run and Evaluate: Execute the Glue job, and review the results for duplicates.

Resolve Duplicates: Export results to an S3 bucket or process them as needed.

This dataset shows severe class imbalance, with only 2% of records representing patients with the disease. AWS ML best practices recommend correcting imbalance only in the training dataset, while keeping validation and test sets representative of real-world distributions.

Synthetic Minority Oversampling Technique (SMOTE) generates synthetic samples of the minority class by interpolating between existing minority examples. This improves the model’s ability to learn disease-related patterns without discarding data.

PCA is a dimensionality reduction method, not an oversampling technique. Oversampling the majority class worsens imbalance. Altering the test dataset would invalidate evaluation results.

Therefore, applying SMOTE to the training dataset is the correct approach.

The text feature contains high-cardinality categorical values with minor spelling variations, which is a common real-world data quality issue. Traditional encoding techniques such as ordinal encoding (Option A) or one-hot encoding (Option C) would treat each misspelled value as a completely separate category, resulting in poor generalization and very high dimensionality.

Similarity encoding addresses this problem by grouping or encoding categories based on string similarity. Categories that differ only slightly—such as spelling errors—are encoded with similar numerical representations. Amazon SageMaker Data Wrangler supports similarity encoding specifically for such noisy categorical features.

Target encoding (Option D) depends on the target label distribution and risks target leakage if not carefully applied.

Therefore, similarity encoding is the most appropriate and AWS-recommended solution.

Managing permissions for multiple Amazon SageMaker notebook instances can become complex when handled individually. To centralize and streamline permission management, AWS recommends creating a single IAM role with the necessary permissions and attaching this role to each notebook instance used by the data science team.

Steps to Implement the Solution:

Create a Single IAM Role with Necessary Permissions:

Define an IAM role that encompasses all permissions required by the data scientists for their tasks. This includes permissions for SageMaker operations and any other AWS services they interact with.

AWS provides managed policies like AmazonSageMakerFullAccess that can be attached to the role to grant comprehensive SageMaker permissions.(IAM Policies for SageMaker)

Attach the IAM Role to Each Notebook Instance:

When creating or updating a SageMaker notebook instance, specify the IAM role created in the previous step. This ensures that all notebook instances operate under a consistent set of permissions.

In the SageMaker console, during the notebook instance setup, you can choose an existing IAM role to associate with the instance.(Creating SageMaker Workspaces)

Benefits of This Approach:

Centralized Permission Management:By using a single IAM role, you simplify the process of updating permissions. Changes to the role ' s policies automatically propagate to all associated notebook instances, ensuring consistent access control.

Adherence to Best Practices:AWS recommends using IAM roles to manage permissions for applications running on services like SageMaker. This approach avoids the need to manage individual user permissions separately.(IAM Best Practices for SageMaker)

Alternative Options and Their Drawbacks:

Option B: Creating a single IAM group and adding data scientists to it does not directly associate the group with notebook instances. IAM groups are used to manage user permissions, not to assign roles to AWS resources like notebook instances.

Option C: Using a single IAM user with the AdministratorAccess policy is not recommended due to security risks associated with granting broad permissions and the challenges in managing shared user credentials.

Option D: Associating an IAM group with a role and then with notebook instances is not a valid approach, as IAM groups cannot be directly associated with AWS resources.

Conclusion: Option A is the most effective solution to centralize and manage permissions for SageMaker notebook instances, aligning with AWS best practices for IAM role management.

Option B is correct because AWS documentation states that speculative decoding is specifically a technique to speed up the decoding process of large language models. The question highlights that the model generates one token at a time and that users are unhappy with the time needed to generate lengthy responses. That is exactly the stage of inference targeted by speculative decoding. AWS explains that this method improves latency without compromising the quality of the generated text.

According to AWS, speculative decoding works by using a smaller, faster draft model to generate candidate tokens first. The larger target model then verifies those tokens. AWS further explains that the draft model can generate multiple candidate tokens quickly, and the target model evaluates them in parallel, which speeds up the final response. This directly addresses the problem in the scenario: the customer support agent is slow not because of startup time, but because long answers require many decoding steps.

The other options are not the best fit. Compilation reduces deployment time and auto-scaling latency through ahead-of-time compilation, but AWS documents it primarily as helping model deployment and hardware optimization rather than token-by-token response generation. Quantization reduces hardware requirements and can lower cost, but the docs do not present it as the most direct answer to slow sequential decoding. Fast model loading improves how quickly the model loads onto instances, which helps startup and scale-out time, not long response generation after the model is already serving traffic. Therefore, the best AWS-documented answer is B.

In Amazon SageMaker AI, identifying bias in machine learning datasets before model training is a critical step to ensure fairness and reliability of predictions. This process is referred to as pre-training bias analysis, and it focuses on understanding whether the training data itself introduces bias—particularly through imbalanced class labels or sensitive attributes.

The Difference in Proportions of Labels (DPL) is a pre-training bias metric specifically designed to measure class imbalance. DPL compares the proportion of a specific label (such as a positive outcome) across different groups or classes within a dataset. If one class or group is overrepresented relative to another, the DPL value will deviate significantly from zero, clearly indicating imbalance. AWS documentation highlights DPL as a key metric used by SageMaker Clarify to detect label imbalance prior to model training.

By contrast, Mean Squared Error (MSE) is a regression evaluation metric used after model training to measure prediction error, not dataset bias. Silhouette score is an unsupervised learning metric used to evaluate clustering quality, making it irrelevant for supervised classification bias detection. Structural Similarity Index Measure (SSIM) is an image-quality metric used in computer vision tasks and has no application in dataset bias analysis.

Using DPL allows ML engineers to proactively detect and address skewed label distributions—such as by re-sampling, re-weighting, or collecting additional data—before training begins. This aligns with AWS best practices for responsible AI and helps reduce the risk of biased predictions that could negatively impact real-world decision-making.

Therefore, Difference in Proportions of Labels (DPL) is the correct and AWS-recommended metric for confirming class imbalance during pre-training bias analysis in Amazon SageMaker AI.

The AWS::SageMaker::Model resource in AWS CloudFormation is used to create an ML model in Amazon SageMaker. This model can then be hosted on an endpoint by using the AWS::SageMaker::Endpoint resource. The model resource defines the container or algorithm to use for hosting and the S3 location of the model artifacts.

Int8 quantization

Low-rank adaptation (LoRA)

Fully Sharded Data Parallel (FSDP)

These three mappings align with AWS documentation for large-model training and optimization in SageMaker AI.

For 1, the correct answer is Int8 quantization. AWS documentation explains that quantization reduces hardware and memory requirements by using a less precise data type for weights and activations, including INT8 formats. AWS Prescriptive Guidance also states that quantization reduces memory footprint by converting higher-precision weights to lower-precision formats such as FP16 to INT8. That matches the description of reducing memory usage by representing model weights with lower precision.

For 2, the correct answer is Low-rank adaptation (LoRA). In practice, LoRA is a standard parameter-efficient fine-tuning (PEFT) method that updates only a small subset of trainable parameters instead of full-model fine-tuning. In SageMaker JumpStart workflows for large language models, LoRA is used specifically to make fine-tuning large models more efficient by training adapter-style low-rank updates rather than all original model weights. This matches the description of a PEFT technique that trains only a subset of model parameters.

For 3, the correct answer is Fully Sharded Data Parallel (FSDP). AWS documentation states that SageMaker model parallelism integrates with PyTorch FSDP, which shards model states across GPUs and supports large-scale distributed training. That directly matches the description of a distributed training technique used to share or shard model parameters across GPUs.

Amazon SageMaker batch transform is ideal for obtaining inferences from large datasets in an asynchronous manner, as it processes data in batches rather than requiring real-time inputs.

SageMaker Model Monitor allows scheduled monitoring of data quality, detecting shifts in input data characteristics, and generating alerts when changes in data quality occur.

This solution provides a fully managed, efficient way to handle both asynchronous inference and data quality monitoring with minimal operational overhead.

Tag model packages and use model package groups that include container images for training and deployment → SageMaker Model Registry

Store predefined language packages, kernels, and relevant dependencies → Amazon Elastic Container Registry (Amazon ECR)

Organize models and their images into model groups for better discoverability → SageMaker Model Registry

Pull built-in SageMaker AI images for model training → Amazon Elastic Container Registry (Amazon ECR)

The correct hotspot mapping is based on the different purposes of SageMaker Model Registry and Amazon ECR.

For model packages, model package groups, and model discoverability, the correct choice is SageMaker Model Registry. AWS documentation states that a model package group is a collection of versioned model packages, and each version can contain the model artifacts, metadata, and container information used for deployment. AWS also documents that registered models can be organized into groups to support governance, discoverability, and lifecycle management. That is why both “tag model packages and use model package groups” and “organize models and their images into model groups” map to SageMaker Model Registry.

For storing software environments and pulling training images, the correct choice is Amazon ECR. AWS documentation says SageMaker uses Docker container images, and these images contain the software stack such as frameworks, language packages, kernels, and dependencies. AWS also states that SageMaker provides prebuilt Docker images for training and inference, and these are referenced through Amazon ECR image URIs. Therefore, both “store predefined language packages, kernels, and relevant dependencies” and “pull built-in SageMaker AI images for model training” map to Amazon ECR.

Amazon Bedrock is a fully managed AWS service and does not reside inside customer VPCs. To allow private connectivity from EC2 instances in a private subnet without using the public internet, AWS documentation recommends AWS PrivateLink.

By creating an interface VPC endpoint for Amazon Bedrock, traffic between the EC2 instances and Bedrock remains entirely within the AWS network. This approach preserves the private subnet configuration and meets strict security requirements.

Security groups alone cannot establish connectivity to a managed service endpoint. Amazon Bedrock cannot be deployed into a customer subnet, making Option C invalid. AWS Direct Connect is designed for on-premises connectivity, not for accessing AWS managed services from within a VPC.

AWS explicitly documents that PrivateLink is the correct mechanism for private access to Amazon Bedrock.

Therefore, Option B is the correct and AWS-verified answer.

Partitioning the time-series data by date prefix in the S3 bucket significantly improves query performance in Amazon Athena by reducing the amount of data that needs to be scanned during queries. This allows the ML engineers to efficiently analyze trends over specific time periods, such as the past 3 days. Applying S3 Lifecycle policies to archive partitions older than 30 days to S3 Glacier Flexible Retrieval ensures cost-effective data retention and storage management while maintaining high performance for recent data retrieval.

The key requirements in this scenario are variable traffic patterns and cost efficiency. The workload has unpredictable spikes during evenings and weekends, followed by long periods of low or no usage. According to AWS Machine Learning documentation, Amazon SageMaker Serverless Inference is specifically designed for such use cases.

SageMaker Serverless Inference automatically provisions, scales, and shuts down compute resources based on incoming inference requests. Customers are billed only for the compute time used during inference, not for idle resources. This makes it highly cost-effective for workloads with intermittent or spiky traffic, such as real-time chat moderation in gaming environments.

Option A is incorrect because batch transform jobs are intended for offline, large-scale inference and require fixed capacity during job execution. They are not suitable for real-time NLP moderation.

Option C is also incorrect because reserving an EC2 GPU instance incurs continuous costs regardless of utilization. This would be inefficient given the long idle periods described in the scenario.

Option D, SageMaker Asynchronous Inference, is designed for workloads with long processing times or large payloads and still requires endpoint provisioning. While it can handle traffic spikes, it does not scale down to zero in the same cost-efficient manner as Serverless Inference.

Therefore, Amazon SageMaker Serverless Inference is the most cost-effective and operationally efficient solution for deploying an NLP moderation model with highly variable usage patterns.

Amazon SageMaker Clarify is the AWS service used to detect and quantify bias and fairness metrics in ML models. When bias monitoring jobs run, Clarify publishes bias metrics directly to Amazon CloudWatch.

CloudWatch metrics can be visualized using CloudWatch dashboards or integrated into other monitoring tools, making them ideal for real-time or periodic bias reporting.

CloudTrail logs API activity and does not capture ML metrics. EventBridge and SNS are used for event routing and notifications, not metric visualization.

AWS documentation explicitly states that Clarify bias metrics are emitted to Amazon CloudWatch, which is the correct source for dashboards.

Therefore, Option B is the correct and AWS-verified answer.

AWS Glue is a serverless data integration service that is well-suited for creating data ingestion pipelines, especially when raw data is stored in Amazon S3. It can clean, transform, and catalog data, making it accessible for downstream ML tasks.

Amazon SageMaker Studio Classic provides a comprehensive environment for building, training, and deploying ML models. It includes built-in tools and capabilities to create efficient model deployment pipelines with minimal setup.

This combination ensures seamless integration of data ingestion and ML model deployment with minimal operational overhead.

AWS Bedrock Knowledge Bases support multiple chunking strategies to optimize retrieval quality. Hierarchical chunking is specifically designed for structured documents such as PDFs with headings, sections, and subsections.

Hierarchical chunking allows fine-grained retrieval at the subsection level while automatically preserving parent section context. This ensures that when a small portion is retrieved, the surrounding section is also provided to the foundation model for better understanding.

Fixed-size and maximum-token chunking can split content arbitrarily, breaking semantic and structural boundaries. Semantic chunking focuses on meaning but does not guarantee structured context preservation without additional logic.

AWS documentation highlights hierarchical chunking as the preferred strategy when documents are structured and contextual integrity is required.

Therefore, Option A is the correct and AWS-aligned solution.

This issue occurs because target tracking auto scaling allows the endpoint to scale down to zero, and scaling up only happens after traffic arrives. At the start of the business day, no instances are running, so the first requests experience cold-start delays.

AWS documentation for Amazon SageMaker recommends using scheduled or step scaling policies when predictable traffic patterns exist. In this case, business hours are predictable, so the best practice is to proactively scale the endpoint before traffic arrives.

Option D correctly uses Amazon CloudWatch alarms with step scaling to increase the minimum instance count from zero to one at the start of the business day. This ensures at least one warm instance is ready to handle requests immediately, eliminating startup latency.

Options A, B, and C do not guarantee instance availability before traffic begins. Cooldown tuning and metric changes only react after load is detected.

Therefore, scheduled step scaling using CloudWatch alarms is the correct solution.

Amazon SageMaker real-time inference endpoints are designed to provide low-latency predictions in production environments. They offer built-in auto scaling to handle unpredictable bursts of requests, ensuring high availability and responsiveness. This approach is fully managed, reduces operational complexity, and is optimized for the range of request sizes (1 KB to 3 MB) specified in the requirements.

Amazon SageMaker AI shadow testing enables ML engineers to evaluate new model versions by sending a copy of live production traffic to a shadow variant without affecting production inference responses. AWS documentation specifies that shadow variants are configured separately from production variants and can use different instance types, including higher-performance GPU instances.

The correct approach is to create a new endpoint configuration using the CreateEndpointConfig API. This configuration includes the existing production variant and a separate ShadowProductionVariants list. The shadow variant can be assigned a larger instance type to meet GPU performance requirements while leaving the production variant unchanged. After creating the configuration, the engineer deploys it using the CreateEndpoint action.

Option A is incorrect because production variant configurations cannot be directly modified to include shadow variants. Option B is incorrect because shadow variants are not defined as standard production variants; defining two production variants would route traffic differently and could affect production behavior. Option C introduces unnecessary complexity and deviates from SageMaker’s built-in shadow testing functionality.

AWS explicitly documents that shadow variants are designed to isolate testing resources, support different instance types, and ensure zero impact on production inference. Therefore, Option D is the correct and AWS-recommended solution.

Amazon Macie is a fully managed data security and privacy service that uses machine learning to discover and classify sensitive data in Amazon S3. It is purpose-built to identify sensitive data with minimal operational overhead. After identifying the sensitive data, you can use AWS Lambda functions to automate the process of removing or redacting the sensitive data, ensuring efficiency and integration with the hybrid cloud environment. This solution requires the least development effort and aligns with the requirement to handle sensitive data effectively.

Text representation of basic units of data processed by LLMs: Token

High-dimensional vectors that contain the semantic meaning of text: Embedding

Enrichment of information from additional data sources to improve a generated response: Retrieval Augmented Generation (RAG)

Comprehensive Detailed Explanation

Token:

Description: A token represents the smallest unit of text (e.g., a word or part of a word) that an LLM processes. For example, " running " might be split into two tokens: " run " and " ing. "

Why? Tokens are the fundamental building blocks for LLM input and output processing, ensuring that the model can understand and generate text efficiently.

Embedding:

Description: High-dimensional vectors that encode the semantic meaning of text. These vectors are representations of words, sentences, or even paragraphs in a way that reflects their relationships and meaning.

Why? Embeddings are essential for enabling similarity search, clustering, or any task requiring semantic understanding. They allow the model to " understand " text contextually.

Retrieval Augmented Generation (RAG):

Description: A technique where information is enriched or retrieved from external data sources (e.g., knowledge bases or document stores) to improve the accuracy and relevance of a model ' s generated responses.

Why? RAG enhances the generative capabilities of LLMs by grounding their responses in factual and up-to-date information, reducing hallucinations in generated text.

By matching these terms to their respective descriptions, the ML engineer can effectively leverage these concepts to build robust and contextually aware generative AI applications on Amazon Bedrock.

Step 1: An S3 event notification invokes the pipeline when new data is uploaded.

Step 2: SageMaker retrains the model by using the data in the S3 bucket.

Step 3: The pipeline deploys the model to a SageMaker endpoint.

Step 1: An S3 Event Notification Invokes the Pipeline When New Data is Uploaded

Why? The CI/CD pipeline should be triggered automatically whenever new training data is uploaded to Amazon S3. S3 event notifications can be configured to send events to AWS services like Lambda, which can then invoke AWS CodePipeline.

How? Configure the S3 bucket to send event notifications (e.g., s3:ObjectCreated:*) to AWS Lambda, which in turn triggers the CodePipeline.

Step 2: SageMaker Retrains the Model by Using the Data in the S3 Bucket

Why? The uploaded data is used to retrain the ML model to incorporate new information and maintain performance. This step is critical to updating the model with fresh data.

How? Define a SageMaker training step in the CI/CD pipeline, which reads the training data from the S3 bucket and retrains the model.

Step 3: The Pipeline Deploys the Model to a SageMaker Endpoint

Why? Once retrained, the updated model must be deployed to a SageMaker endpoint to make it available for real-time inference.

How? Add a deployment step in the CI/CD pipeline, which automates the creation or update of the SageMaker endpoint with the retrained model.

Order Summary:

An S3 event notification invokes the pipeline when new data is uploaded.

SageMaker retrains the model by using the data in the S3 bucket.

The pipeline deploys the model to a SageMaker endpoint.

This configuration ensures an automated, efficient, and scalable CI/CD pipeline for continuous retraining and deployment of the ML model in Amazon SageMaker.

In distributed training and processing jobs, multiple instances and containers communicate with each other over the network to exchange gradients, parameters, and intermediate results. Even when jobs run inside a private VPC, network traffic between instances is not automatically encrypted at the application layer.

AWS documentation for Amazon SageMaker specifies that inter-container traffic encryption is the supported mechanism for encrypting data in transit between containers that participate in distributed training or processing jobs. When enabled, SageMaker uses TLS to encrypt all communication between containers across instances, ensuring confidentiality and compliance with security requirements.

Option A (network isolation) prevents containers from making outbound network calls but does not encrypt traffic between distributed instances. Option B is incorrect because security groups control traffic access, not encryption. Option D (VPC flow logs) is a monitoring feature and does not provide encryption.

Therefore, enabling inter-container traffic encryption is the correct and AWS-recommended solution.

Recall measures the ability of a model to correctly identify all positive cases (true positives) out of all actual positives, minimizing false negatives. Since the cost of false negatives is much higher than false positives in this scenario, the ML engineer should prioritize models with high recall to reduce the likelihood of missing positive cases.

This scenario describes a severely imbalanced classification problem. In such cases, accuracy is a misleading metric, because the model can achieve high accuracy by predicting only the majority class.

AWS ML best practices recommend using F1 score (or precision/recall) when evaluating imbalanced datasets. The F1 score balances false positives and false negatives, making it ideal for assessing minority-class performance.

For image data, image augmentation (rotations, flips, crops, color jitter) is the preferred technique to increase minority-class representation. SMOTE is designed for tabular data and is not suitable for image pixel data.

Therefore, the correct solution is to optimize for F1 score and apply image augmentation.

Thus, Option B is the correct and AWS-aligned answer.

Option C is correct because the requirement has two separate parts: first, the company must track model versions; second, it needs recommendations for which instance types to use for hosting the model in SageMaker AI. AWS documentation identifies the SageMaker Model Registry as the SageMaker feature used to register and manage model packages and versions as part of the ML lifecycle. For hosting recommendations, AWS documentation says Amazon SageMaker Inference Recommender helps select the best instance type and configuration for ML models and workloads by automating benchmarking and load testing across SageMaker AI instances.

The AWS Inference Recommender documentation is especially important here because it explicitly says you can use it after you register a model to the SageMaker Model Registry with model artifacts. It also states that Inference Recommender helps choose the best endpoint type and configuration and returns recommendations that include the instance type in the resulting endpoint configuration. That is an exact match to the question’s requirement to recommend hosting instance types for a model that has not yet been hosted on SageMaker AI.

The other options do not match AWS service purposes. Amazon ECR stores container images, not model versions. AWS Compute Optimizer is not the SageMaker-native service documented for inference benchmarking of ML models. SageMaker Autopilot is for automated model building, not inference instance recommendations. SageMaker Experiments tracks experiment metadata, trials, and runs, but it does not recommend deployment instance types. Therefore, the fully verified AWS-docs answer is C.

In regression problems such as house price prediction, extreme values can significantly distort model learning. In this dataset, the presence of a large mansion and an extremely small apartment represents clear outliers in the Square Meters feature. According to AWS Machine Learning best practices, outliers can disproportionately influence loss functions (such as mean squared error), leading to poor predictions for the majority of typical data points.

Removing these outliers helps the model focus on learning patterns that apply to the majority of houses and apartments, which aligns with the requirement to produce accurate predictions for typical properties. After removing outliers, applying a log transformation to the Square Meters feature further improves model performance by reducing skewness and stabilizing variance. Log transformations are commonly recommended in AWS and general ML documentation when numerical features span multiple orders of magnitude.

Option B is incorrect because normalization alone does not address the undue influence of extreme outliers. Option C and D are incorrect because one-hot encoding is intended for categorical variables, not continuous numerical features such as square meters.

Therefore, removing outliers and applying a log transformation is the most statistically sound preprocessing approach.

AWS recommends SageMaker shadow testing as the safest way to validate a new model version using real production traffic without impacting production responses. A shadow variant receives a copy of inference requests but does not return predictions to end users. This completely eliminates the risk of exposing incorrect predictions.

Options A and B are canary deployments. While useful, they still allow v2 to return responses to real users, which violates the requirement to prevent disruption. Option C sends 100% of traffic to v2 externally and is risky.

Shadow variants are explicitly designed to minimize risk while using real data, making them the AWS best practice for pre-production validation.

Therefore, Option D is correct.

AWS Secrets Manager is designed for securely storing, managing, and automatically rotating secrets, including API tokens. By configuring a Lambda function for custom rotation logic, the solution can automatically rotate the API tokens every 3 months as required. Secrets Manager simplifies secret management and integrates seamlessly with other AWS services, making it the ideal choice for this use case.

Linear regression model whose coefficients should shrink but not become zero

Answer: L2 regularization

AWS says L2 produces smaller overall weight values and is the right fit when coefficients should be reduced without being forced to zero.

Polynomial regression model with irrelevant polynomial terms that should be eliminated

Answer: L1 regularization

AWS says L1 reduces the number of features used by pushing small weights to zero, which matches elimination of irrelevant terms.

Logistic regression model that has highly correlated features to eliminate highly redundant predictors

Answer: L1 regularization

This is the nuanced one. AWS says L2 stabilizes weights when there is high correlation between features, but because the question explicitly says eliminate highly redundant predictors, L1 is the better match since it creates sparsity and removes predictors by zeroing coefficients.

Amazon SageMaker Data Wrangler is a comprehensive tool that streamlines the process of data preparation and offers built-in capabilities for anomaly detection and visualization.

Key Features of SageMaker Data Wrangler:

Data Importation: Connects seamlessly to various data sources, including Amazon S3 and on-premises databases, facilitating the aggregation of transaction logs, customer profiles, and MySQL tables.

Anomaly Detection: Provides built-in analyses to detect anomalies in time series data, enabling the identification of outliers that may indicate fraudulent activities.

Visualization: Offers a suite of visualization tools, such as histograms and scatter plots, to help understand data distributions and relationships, which are crucial for feature engineering and model development.

Implementation Steps:

Data Aggregation:

Import data from Amazon S3 and on-premises MySQL databases into SageMaker Data Wrangler.

Utilize Data Wrangler ' s data flow interface to combine and preprocess datasets, ensuring a unified dataset for analysis.

Anomaly Detection:

Apply the anomaly detection analysis feature to identify outliers in the dataset.

Configure parameters such as the anomaly threshold to fine-tune the detection sensitivity.

Visualization:

Use built-in visualization tools to create charts and graphs that depict data distributions and highlight anomalies.

Interpret these visualizations to gain insights into potential fraud patterns and feature interdependencies.

Advantages of Using SageMaker Data Wrangler:

Integrated Workflow: Combines data preparation, anomaly detection, and visualization within a single interface, streamlining the ML development process.

Operational Efficiency: Reduces the need for multiple tools and complex integrations, thereby minimizing operational overhead.

Scalability: Handles large datasets efficiently, making it suitable for extensive transaction logs and customer profiles.

By leveraging SageMaker Data Wrangler, the ML engineer can effectively detect anomalies and visualize results, facilitating the development of a robust fraud detection model.

Analyze and Visualize - Amazon SageMaker

Transform Data - Amazon SageMaker

SageMaker Serverless Inference is ideal for workloads with predictable, intermittent demand. By enabling provisioned concurrency, the model can handle multiple invocations quickly during the high-demand 2-hour period. AWS manages the underlying infrastructure and scaling, ensuring the solution meets performance requirements with minimal operational overhead. This approach is cost-effective since it scales down when not in use.

Option D is correct because Amazon Bedrock Knowledge Bases are designed for applications that need to answer questions using private documents without retraining or repeatedly fine-tuning a foundation model. AWS documentation states that with Amazon Bedrock Knowledge Bases, you can build applications enriched by context retrieved from a knowledge base, and that this provides an out-of-the-box RAG solution. AWS also explicitly says that adding a knowledge base increases cost-effectiveness by removing the need to continually train your model to use your private data. That matches this use case very closely.

The question also says the documentation consists of text files that are only several megabytes total and are updated often. A retrieval-based approach is more economical and operationally simpler than creating a new model or repeatedly fine-tuning one whenever the documents change. AWS documentation for Bedrock knowledge bases describes adding data sources and running ingestion jobs to process and index the content, which is exactly the pattern needed for frequently updated internal documentation used by a chat interface.

The other options are not as cost-effective. Creating a new LLM is far beyond the need here. Guardrails help control model behavior and policy enforcement, but they do not serve as a document retrieval layer for internal documentation. Fine-tuning a model on frequently changing text files is usually more expensive and less flexible than using retrieval augmentation. For a modest-sized, frequently updated documentation corpus, the AWS-native and most cost-effective solution is to load the files into an Amazon Bedrock knowledge base and use it to provide context at inference time. Therefore, the best verified answer is D.

Amazon Kendra Experience Builder provides a fully managed, low-code solution for building conversational search and question-answering applications. AWS documentation states that Kendra natively supports semantic search, vector embeddings, and vector-based retrieval, making it well suited for RAG-style applications with minimal development effort.

When integrated with Amazon Bedrock, Kendra can act as the retrieval layer, handling document ingestion, indexing, embedding generation, and relevance ranking automatically. This eliminates the need to manually manage embedding models, vector databases, and search logic.

Options B and C require custom schema design, vector indexing, query logic, and operational management of PostgreSQL instances. Although pgvector supports vector search, it significantly increases development and maintenance effort. Option D is unrelated to vector search and is used only for metadata cataloging.

AWS explicitly positions Amazon Kendra as the fastest way to build enterprise-grade conversational assistants that integrate with foundation models.

Therefore, Option A is the correct and most AWS-aligned solution.

This problem describes a recommendation system with millions of records, many categorical variables, and a high-dimensional sparse feature space created by one-hot encoding. AWS documentation explicitly recommends Amazon SageMaker Factorization Machines (FM) for such use cases.

Factorization Machines are designed to handle sparse datasets efficiently and to model interactions between categorical features without explicitly enumerating all feature combinations. This capability makes FM particularly well-suited for recommendation problems such as predicting user-item interactions, including destination recommendations.

With 2,000 possible airport destinations, the target space is large and sparse. One-hot encoding further increases sparsity. Factorization Machines address this challenge by learning latent factors that capture relationships between features, even when many feature combinations are rarely observed.

Option A (CatBoost) is not an Amazon SageMaker built-in algorithm and therefore does not meet the requirement. Option B (DeepAR) is a time-series forecasting algorithm, not intended for recommendation or classification problems. Option D (k-means) is an unsupervised clustering algorithm and cannot directly predict a specific destination label.

AWS documentation explicitly lists recommendation systems and click prediction as primary use cases for the SageMaker Factorization Machines algorithm.

Therefore, Option C is the correct and AWS-verified choice.

Amazon Comprehend provides the ImportModel API operation, which allows you to copy a custom model between AWS accounts. By creating a resource-based IAM policy on the model in Account A, you can grant Account B the necessary permissions to access and import the model. This approach requires minimal development effort and is the AWS-recommended method for sharing custom models across accounts.

Option A is correct because AWS Glue is the AWS-native managed ETL service built specifically to discover schema, run ETL jobs, and store metadata in the AWS Glue Data Catalog. AWS documentation states that Glue crawlers can automatically discover and catalog new or updated data sources, and that the Data Catalog automatically captures and manages schema metadata. This directly matches the requirement to run an ETL job on data in Amazon S3, discover the schema, and store the metadata with the least manual effort.

AWS Glue is also the lowest-effort answer because the service is managed and purpose-built for this workflow. The Glue Data Catalog serves as a persistent metadata repository, and AWS documents that crawlers infer schema information and integrate it into the catalog automatically. That means the ML engineer does not need to build custom schema inference logic or manually maintain metadata storage. This is exactly the kind of manual work the question is trying to avoid.

The other options are not as good. SageMaker Data Wrangler is primarily for visual data preparation and feature engineering, not for running a managed ETL-plus-catalog workflow with schema stored in a metadata catalog. Athena with Step Functions would require assembling more custom orchestration and still does not naturally replace the Glue Data Catalog workflow. Launching an EC2 instance introduces the highest operational overhead and does not align with the requirement for least manual effort. Therefore, the best verified AWS-docs answer is A, because AWS Glue combines ETL, schema discovery, and metadata cataloging in one managed service.

Bias detection is a critical step in responsible machine learning and is emphasized in AWS documentation, particularly in Amazon SageMaker Clarify. When analyzing structured datasets that include sensitive or influential attributes such as age, AWS recommends evaluating label distribution fairness and group-based outcome differences before training a model.

The class imbalance metric helps identify whether certain outcomes (for example, purchase vs. no purchase) are overrepresented or underrepresented. Severe imbalance can cause models to favor majority classes, leading to biased predictions. AWS explicitly highlights class imbalance as a key issue to assess during data exploration.

The Difference in Proportions of Labels (DPL) is a fairness metric supported by SageMaker Clarify that measures whether outcome labels are disproportionately distributed across different groups, such as age ranges. DPL compares the proportion of favorable outcomes between groups, making it especially effective for identifying demographic bias in categorical labels.

Options A and D focus on descriptive statistics or correlations, which are useful for data understanding but do not directly measure bias or fairness. Option B partially addresses imbalance and sentiment but sentiment analysis of reviews alone does not quantify demographic bias tied to outcomes.

AWS documentation strongly recommends using group fairness metrics, including DPL, alongside class imbalance checks to identify bias before training. These metrics provide actionable insights into whether a dataset may lead to unfair or skewed model behavior.

Therefore, Option C is the most appropriate and AWS-aligned choice.

Amazon SageMaker automatically tracks the Invocations metric, which represents the number of API calls made to the endpoint, in Amazon CloudWatch. By adding this metric to a CloudWatch dashboard, you can monitor the endpoint ' s activity in real-time. Setting up a CloudWatch alarm allows the system to send notifications whenever the API call events exceed the defined threshold, meeting both the monitoring and notification requirements efficiently.

AWS documentation strongly recommends using custom Docker containers when ML workloads require consistent access to custom dependencies across processing jobs, training jobs, and pipelines.

By building a single Docker image that contains all required libraries and hosting it in Amazon ECR, the ML engineer ensures that every SageMaker job uses the same runtime environment. This approach eliminates the need for repetitive installation steps and avoids environment drift.

Manually installing libraries in managed containers is error-prone and not reusable across jobs. Notebook instances are not designed to host production jobs and pipelines. Running code externally breaks the SageMaker workflow and increases operational complexity.

Using a custom container is a one-time setup that provides maximum reuse with minimal ongoing effort, making it the least implementation effort option in the long run.

Therefore, Option B is the correct and AWS-recommended answer.

The company lacks programming expertise, so a no-code/low-code solution is required. Amazon SageMaker Canvas includes Data Wrangler’s visual interface, which enables users to import, transform, and prepare data using a graphical workflow.

Data Wrangler in Canvas supports common preprocessing tasks—such as handling missing values, normalization, outlier detection, and feature engineering—without writing code. It integrates directly with Amazon S3 and is suitable for large tabular datasets.

Options A, C, and D require coding skills (Spark, SQL, or Python), which violates the stated constraint.

Therefore, using the visual interface of SageMaker Data Wrangler in Canvas is the correct solution.

Network ACLs (Access Control Lists) operate at the subnet level and allow for rules to explicitly deny traffic from specific IP addresses. By creating an inbound rule in the network ACL to deny traffic from the suspicious IP address, the company can block traffic to the Amazon SageMaker domain from that IP. This approach works because network ACLs are evaluated before traffic reaches the security groups, making them effective for blocking traffic at the subnet level.

The day_of_week feature is a categorical variable with a small, fixed number of unique values and no inherent ordinal relationship. AWS machine learning best practices strongly recommend one-hot encoding for this type of categorical data when preparing features for classification models.

One-hot encoding converts each unique category into a separate binary feature (0 or 1). For example, “Monday” becomes a column where Monday = 1 and all other days = 0. This ensures that the ML model does not incorrectly assume a numeric or ordered relationship between categories.

Option B (label encoding) assigns integer values to categories (e.g., Monday = 1, Tuesday = 2). AWS documentation cautions against this approach for nominal data because models may incorrectly infer ordinal meaning, leading to biased or inaccurate predictions.

Option A (binary encoding) is typically used for high-cardinality categorical features to reduce dimensionality. With only seven categories, AWS recommends one-hot encoding for clarity and interpretability.

Option D (tokenization) is used for text processing, such as NLP tasks, and is not appropriate for structured categorical features.

AWS SageMaker feature engineering guidelines emphasize that one-hot encoding is the preferred method for low-cardinality categorical variables in classification models, especially when using algorithms such as logistic regression, neural networks, and tree-based models.

Therefore, Option C is the correct and AWS-aligned choice.

Problem Description:

The F1 score, which is a balance of precision and recall, has decreased significantly. This indicates the model ' s predictions are no longer aligned with the real-world data distribution.

Why Concept Drift?

Concept drift occurs when the statistical properties of the target variable or features change over time. For example, customer behaviors or subscription cancellation patterns may have shifted, leading to reduced model accuracy.

Signs of Concept Drift:

Deviation in performance metrics (e.g., F1 score) over time.

Declining prediction accuracy for certain groups or scenarios.

Solution:

Monitor for drift using tools like SageMaker Model Monitor.

Regularly retrain the model with updated data to account for the drift.

Why Not Other Options?:

B: Model complexity is unrelated if the model initially performed well.

C: Data quality issues would have been detected during baseline analysis.

D: Incorrect ground truth labels would have resulted in a consistently poor baseline.

Conclusion: The decrease in F1 score is most likely due to concept drift in the customer data, requiring retraining of the model with new data.

The requirement is to quickly determine whether the target variable is predictable, with minimal development effort. Amazon SageMaker Autopilot is specifically designed for this purpose.

SageMaker Autopilot automatically handles data preprocessing, feature engineering, algorithm selection, model training, and evaluation. It generates multiple candidate models and provides detailed performance metrics, allowing teams to quickly assess predictability without writing custom code.

Option B requires significant manual development. Option C is incorrect because Amazon Macie is a data security and classification service, not an ML modeling service. Option D is unsuitable because Amazon Bedrock models are not intended for structured tabular prediction tasks.

Therefore, SageMaker Autopilot provides the fastest and least effort solution.

For real-time video content moderation with minimal operational overhead, AWS documentation recommends using fully managed, purpose-built AI services. Amazon Rekognition provides real-time video analysis capabilities, including content moderation, unsafe content detection, and label recognition for live video streams.

By integrating Rekognition with AWS Lambda, the company can automatically process video frames, extract moderation metadata, and take immediate action (such as flagging or stopping a stream) without managing servers, models, or infrastructure. This serverless architecture scales automatically and minimizes operational complexity.

Option B introduces unnecessary complexity. While Amazon Bedrock LLMs are powerful, they are not required for image-based moderation tasks that Rekognition already handles natively.

Option C is incorrect because using Amazon SageMaker would require model training, endpoint management, and scaling, significantly increasing operational overhead.

Option D is incorrect because Amazon Transcribe and Amazon Comprehend are designed for audio and text analysis, not image or video frame moderation.

Therefore, Amazon Rekognition with AWS Lambda is the most efficient, scalable, and low-maintenance solution for real-time video moderation during live streaming events.

A serverless inference endpoint in Amazon SageMaker is ideal for use cases where the model is invoked infrequently, such as running one time each night. It eliminates the cost of idle resources when the model is not in use. Setting the MaxConcurrency parameter to 1 ensures cost-efficiency while supporting the required single nightly invocation. This solution minimizes costs and matches the requirement to process a small amount of data quickly.

SageMaker Debugger can identify when a training job is not converging or is stuck in a non-productive state. By stopping these jobs early, unnecessary energy and computational resources are conserved, improving sustainability.

AWS Trainium instances are purpose-built for ML training and are optimized for energy efficiency and cost-effectiveness. They use less energy per training task compared to general-purpose instances, making them a sustainable choice.

Option A is correct because Amazon SageMaker Model Monitor is the AWS service specifically built to monitor data quality and model quality for ML models in production. AWS documentation states that Model Monitor supports continuous monitoring with a batch transform job that runs regularly and also supports on-schedule monitoring for asynchronous batch transform jobs. That aligns directly with the scenario of a hospital running a nightly batch inference job and needing a daily report on both the incoming data and the model’s predictive performance.

AWS documentation also separates the two monitoring needs very clearly. Data quality monitoring can be scheduled for batch transform jobs by using DefaultModelMonitor with a BatchTransformInput. Model quality monitoring can also be scheduled for batch transform jobs by using ModelQualityMonitor, which compares predictions against actual ground-truth labels stored in Amazon S3. Since the question explicitly asks for both model data quality and model performance, SageMaker Model Monitor is the documented feature that covers both requirements together.

Option B is not sufficient because CloudWatch dashboards show operational and resource metrics, not the full ML-specific data quality and model quality reports required here. Option C is incorrect because AWS Glue DataBrew is for data preparation and profiling, not model performance monitoring. Option D is partially plausible because SageMaker Pipelines integrates with QualityCheck steps and can run monitoring jobs on demand, but the AWS docs position Model Monitor as the native solution for scheduled monitoring of production batch inference workloads. Therefore, the best AWS-documented answer is A.

Amazon SageMaker Data Wrangler provides a comprehensive solution for importing, consolidating, and preparing datasets for ML. It offers tools to handle missing values, duplicates, and outliers through its built-in cleansing and enrichment functionalities, allowing the ML engineer to efficiently prepare the data in a single environment with minimal manual effort.

The requirement that each parameter trial is built based on the performance of the previous trial is the defining characteristic of Bayesian optimization. In Amazon SageMaker Automatic Model Tuning, Bayesian optimization uses prior trial results to intelligently select the next set of hyperparameters, making it far more efficient than grid or random search—especially in large, mixed search spaces.

This scenario includes both categorical (20) and continuous (7) hyperparameters and allows up to 1,000 training jobs, which is well within the supported limits of SageMaker AMT. Bayesian optimization natively supports mixed parameter types and is explicitly recommended by AWS for large, high-dimensional search spaces where exhaustive grid search is impractical.

Option A and D (grid search) do not meet the requirement because grid search evaluates combinations independently and does not learn from previous trials. Additionally, grid search becomes computationally infeasible as dimensionality increases.

Option B (random search) also evaluates trials independently and does not leverage previous results, violating the core requirement.

Therefore, defining both categorical and continuous parameters and using Bayesian optimization with a maximum of 1,000 jobs is the correct and AWS-recommended solution.

Amazon MSK Serverless provides a fully managed Apache Kafka-compatible service that automatically handles provisioning, scaling, and capacity management. AWS documentation states that MSK Serverless is designed for customers who want Kafka functionality without managing infrastructure.

Option B requires capacity planning and scaling management. Option C uses proprietary technology rather than open source. Option D requires full infrastructure management.

MSK Serverless delivers near real-time streaming with minimal operational overhead while maintaining compatibility with open-source Kafka tooling.

Therefore, Option A is the correct solution.

Option A is correct because Amazon SageMaker Feature Store is the AWS service designed to act as a centralized repository for ML features that are used consistently across training and inference. AWS documentation states that SageMaker Feature Store simplifies how you create, store, share, and manage features for data exploration, model training, and model inference. This directly matches the requirement to create a repository for streamed song ratings that will be used in both batch training and real-time inference.

The most important requirement in the question is that the data must stay synchronized between batch training and real-time inference. AWS documents explain that Feature Store provides both an offline store and an online store. The offline store is used for historical data, model training, and batch inference, while the online store is a low-latency, high-availability store intended for real-time lookup during inference. This dual-store design is exactly why Feature Store is used to maintain feature consistency across training and serving workflows. AWS Well-Architected guidance also explicitly says Feature Store provides online storage for real-time inference and offline storage for model training and batch inference.

The other options do not solve the full problem. Athena CTAS can organize query results but does not provide a synchronized feature repository for online and offline ML use. Lake Formation governs access to data lakes but is not a feature repository for training and inference consistency. Data Wrangler Generate Data Insights is for analysis and preparation, not synchronized feature serving. Therefore, the best AWS-documented answer is A.

Amazon SageMaker Pipelines is purpose-built for orchestrating end-to-end ML workflows, including data ingestion, training, evaluation, and deployment. It supports automation, versioning, and deployment of the latest model versions.

MWAA orchestrates general workflows but lacks ML-native features. Lambda cannot handle long-running ML training. Spark processes data but does not manage ML lifecycle.

Therefore, Option A is the correct AWS-native solution.

For cross-account Amazon S3 access, AWS requires two components:

An S3 bucket policy in the owning account (Account A) that grants access to a principal in another account

An IAM role policy in the consuming account (Account B) that allows the service to access the bucket

Amazon SageMaker training jobs assume an IAM role in the account where the job runs—in this case, Account B. Therefore, the IAM policy must be attached to the SageMaker execution role in Account B.

The S3 bucket policy must reside in Account A because bucket policies are owned and enforced by the bucket owner. This policy explicitly allows the IAM role from Account B to read the training data.

Any other combination fails either because the policy is in the wrong account or because the role is not the one used by SageMaker.

AWS documentation clearly describes this pattern as the correct way to grant cross-account access for SageMaker training jobs.

Therefore, Option B is the correct and AWS-aligned solution.

The key requirements are real-time processing, high throughput, and minimal operational overhead. Amazon Kinesis Data Streams is designed for ingesting thousands of events per second with low latency.

For anomaly detection on streaming data, Amazon Managed Service for Apache Flink provides a built-in Random Cut Forest (RCF) function. RCF is an unsupervised anomaly detection algorithm that works well on numerical streaming data and does not require labeled training data.

This fully managed combination eliminates the need to deploy or maintain SageMaker endpoints, EC2 instances, or custom ML pipelines. Options B and C introduce unnecessary infrastructure and model management overhead. Option D is batch-oriented and unsuitable for real-time anomaly detection.

Therefore, using Kinesis Data Streams with Flink’s built-in Random Cut Forest is the most scalable and low-overhead solution.

AWS Glue DataBrew supports datasets sourced directly from Amazon S3 paths. To clean and normalize data, AWS documentation specifies using DataBrew recipes, which define transformation steps such as standardization, deduplication, and formatting.

A profile job is used only for data analysis and statistics generation, not for transformation. A recipe job applies actual transformations to the data and writes the output to Amazon S3.

JDBC connections are used for relational databases, not S3 buckets. Therefore, options C and D are invalid.

AWS clearly documents that the correct workflow is:

Create a DataBrew dataset from an S3 path

Define transformations using a recipe

Run a recipe job to produce cleaned data

Thus, Option B is the correct and AWS-aligned answer.

Amazon Kendra is an AI-powered search service designed for semantic search use cases. It allows ingestion of documents from an Amazon S3 bucket using the Amazon Kendra S3 connector. Once the documents are ingested, Kendra enables semantic searches with its built-in capabilities, removing the need to manually generate embeddings or manage a vector database. This approach is efficient, requires minimal operational effort, and meets the requirements for a Retrieval Augmented Generation (RAG) application.

Logging the metrics to Amazon CloudWatch allows the metrics to be tracked and monitored effectively.

CloudWatch Alarms can be configured to trigger when metrics breach a predefined threshold.

The alarm can be set to notify through Amazon Simple Notification Service (SNS), which can send email messages to the configured recipients.

This is the standard and most efficient way to achieve the desired functionality.

Apache Parquet is a columnar storage file format optimized for complex and large datasets. It provides efficient reading and processing by accessing only the required columns, which reduces I/O and speeds up data handling. This makes it ideal for use with Amazon SageMaker Canvas, where minimizing processing time is important for training ML models. Parquet is also compatible with S3 and widely supported in data analytics and ML workflows.

Step 1:

Create a Shapley Additive Explanations (SHAP) baseline for the model by using Amazon SageMaker Clarify.

Step 2:

Schedule an hourly model explainability monitor.

Step 3:

Check the analysis results on the SageMaker Studio console.

When monitoring a deployed model for data drift and explainability, AWS prescribes a specific workflow using SageMaker Clarify and SageMaker Model Monitor:

Create a SHAP baseline (Step 1)Before any monitoring can occur, SageMaker Clarify must establish a baseline explainability configuration. This baseline captures the reference SHAP values for feature importance using training or baseline data. Model Monitor uses this baseline to compare future inferences and detect drift in feature attributions.

Schedule the model explainability monitor (Step 2)After the baseline is created, an explainability monitoring schedule must be configured (hourly in this case). The monitor periodically analyzes inference data, compares it against the SHAP baseline, and generates reports that highlight drift or anomalies in feature contributions.

Inspect results in SageMaker Studio (Step 3)Once monitoring jobs run, SageMaker stores the analysis results in Amazon S3 and surfaces them in the SageMaker Studio console, where engineers can review metrics, violations, and visual reports.

This sequence is mandatory because:

A monitor cannot run without a baseline

Results cannot be reviewed until the monitor executes

Option B is correct because AWS Glue provides a built-in Detect PII transform that can detect, mask, or remove personally identifiable information with minimal custom development. AWS documentation says the Detect PII transform can process predefined AWS-managed PII entity types and supports actions such as removing or masking values. The examples in AWS docs explicitly mention sensitive entities such as phone numbers and addresses, which directly match the problem statement.

The question specifically asks for the least development effort. That wording makes AWS Glue Detect PII the strongest answer because it is a native transformation capability rather than a custom code-heavy workflow. AWS also documents fine-grained sensitive data detection features that let you apply actions per entity type, improving usability and reducing the need to build custom parsing and redaction logic yourself. This is much easier than creating Lambda-based transformation code or custom text-cleaning logic inside another ML preprocessing tool.

The other options are less suitable. Amazon Bedrock Guardrails is not the standard AWS service documented for bulk ETL-style redaction of training text files in this context. S3 Object Lambda would require more custom engineering to inspect and redact each object. SageMaker Data Wrangler custom transformation would also involve extra implementation work compared with using a purpose-built Glue transform. Because the call center already has text output and simply needs regulated fields like names, addresses, and phone numbers removed before training, the AWS-native low-effort solution is AWS Glue Detect PII. Therefore, the best verified answer is B.

SageMaker Data Wrangler provides a no-code/low-code interface for preparing and transforming data, including dropping unnecessary columns. By creating a data flow and configuring a transform step, the ML engineer can easily remove correlated or unneeded columns from the Parquet file with minimal effort. This approach avoids the need for custom coding or managing additional infrastructure.

An oscillating loss pattern during training with stochastic gradient descent (SGD) is a strong indicator that the learning rate is too high. When the learning rate is excessive, the optimizer takes overly large steps during gradient updates, causing the model to repeatedly overshoot the optimal minimum of the loss function. This results in unstable convergence behavior, where training and validation loss decrease briefly and then increase again in a repeating cycle.

AWS Machine Learning documentation and general deep learning best practices recommend reducing the learning rate when training loss and validation loss both remain high and fluctuate rather than steadily decreasing. Lowering the learning rate allows the optimizer to take smaller, more precise steps toward the minimum, leading to smoother convergence and improved generalization on the test dataset.

Option A, early stopping, is used primarily to prevent overfitting when validation loss increases while training loss continues to decrease. In this scenario, both losses remain high and unstable, indicating an optimization issue rather than overfitting.

Option B is incorrect because increasing the test set size does not affect the training dynamics or convergence behavior of the model.

Option C would worsen the problem, as increasing the learning rate would further amplify oscillations and instability.

Therefore, decreasing the learning rate is the correct corrective action to stabilize SGD training and improve model performance.

Amazon SageMaker Asynchronous Inference is specifically designed for long-running inference requests and large payloads. It supports payload sizes up to 1 GB and processing times of up to 1 hour, while automatically queuing requests.

Asynchronous inference stores results in Amazon S3 and allows clients to retrieve inference outputs after processing completes. It also supports auto scaling down to zero when there are no incoming requests, reducing cost.

Batch transform is intended for offline, bulk inference and does not return per-request results in an asynchronous request–response pattern. Serverless and real-time inference have strict payload size and timeout limits that do not support 1-hour processing.

Therefore, asynchronous inference is the only SageMaker inference option that meets all stated requirements.