AWS · MLA-C01 · Intermediate

AWS Machine Learning Engineer - Associate (MLA-C01) — Practice Questions and Mock Exam

Name: Pass-IT
Author: Pass-IT

Prepare for MLA-C01 with original practice questions and clear answer explanations. Ask Alex, your AI tutor, when you need more detail, use your results to identify topics to review, and practice your pacing with timed mock exams.

Start practicing MLA-C01 free Buy MLA-C01 access for $29.99

Buy MLA-C01 access for $29.99 Start practicing MLA-C01 free

65Mock exam questions

130minTime limit

720/ 1000Passing score

Checked against AWS · August 2026 · Current exam version

Overview

About the exam

The AWS Certified Machine Learning Engineer – Associate validates the ability to build, train, deploy, and maintain machine learning models in production on AWS. It covers data preparation for ML, model development and tuning, deployment and orchestration of ML workflows, and monitoring and securing ML solutions using services like SageMaker.

This certification is designed for ML engineers, data scientists, and MLOps practitioners with at least one year of experience using AWS ML services. It demonstrates proficiency in operationalizing ML models, building automated training pipelines, and implementing responsible AI practices in production environments.

Try the practice preview

Try five MLA-C01 questions

Try five practice questions from the app’s current AWS Certified Machine Learning Engineer - Associate question bank, with answers and explanations.

Deployment and Orchestration of ML Workflows1 / 5

An ML engineer is deploying a computer vision model that requires GPU inference. The model receives steady traffic during business hours but minimal traffic overnight. Which SageMaker deployment configuration optimizes cost while ensuring GPU availability during business hours?

AlexFull explanation from Alex

Scheduled auto scaling allows you to proactively set endpoint capacity using cron-like expressions that match known traffic patterns (AWS Auto Scaling docs). For GPU workloads with predictable business-hours traffic, this avoids running expensive GPU instances 24/7 while ensuring capacity is pre-scaled before demand arrives. Serverless Inference does not support GPU instance types, making it unsuitable for GPU workloads. CPU instances lack the processing power required for computer vision inference, resulting in unacceptable latency. Batch Transform processes data in batches rather than serving real-time predictions, so it cannot handle incoming requests on demand.

Sourcedocs.aws.amazon.com

ML Model Development2 / 5

An ML engineer needs to fine-tune a pre-trained large language model using a small labeled dataset. They want to minimize catastrophic forgetting of the model's general knowledge. Which technique is most appropriate?

AlexFull explanation from Alex

PEFT methods like LoRA (Low-Rank Adaptation) freeze the original pre-trained weights and only train small adapter matrices, preserving the model's general knowledge while adapting to new tasks (SageMaker JumpStart fine-tuning docs). This directly addresses catastrophic forgetting by minimizing parameter updates. Increasing training epochs to 100+ worsens catastrophic forgetting by over-training on the small dataset. Reinitializing all model weights destroys the pre-trained knowledge entirely, defeating the purpose of fine-tuning. Training all layers with a high learning rate causes aggressive weight updates that overwrite learned representations, maximizing catastrophic forgetting.

Sourceaws.amazon.com

Data Preparation for Machine Learning (ML)3 / 5

An ML engineer is using AWS Glue crawlers to catalog data stored in Amazon S3 for ML training. The crawler discovers a new table, but the inferred schema has incorrect column data types. What should the engineer do to fix the existing table definition with the least operational effort?

AlexFull explanation from Alex

AWS Glue crawlers infer schemas by using built-in or custom classifiers, and those classifiers can affect the schema written to the Data Catalog. After a table exists, the Data Catalog schema can also be edited directly. For a one-off incorrect inferred data type, manually updating the table schema is the most direct fix; for repeated inference errors, consider adding or adjusting a custom classifier and crawler behavior.

Sourcedocs.aws.amazon.com

ML Model Development5 / 5

An ML engineer is training a deep learning model on Amazon SageMaker. The engineer notices that the training loss continues to decrease but the validation loss starts increasing after epoch 15. Which technique should the ML engineer apply to address this issue?

AlexFull explanation from Alex

When training loss decreases but validation loss increases, the model is overfitting — memorizing training data rather than learning generalizable patterns (AWS ML overfitting docs). Early stopping monitors validation metrics and halts training when improvement stops, preserving the best-performing weights. This is the simplest and most effective remedy. Increasing epochs to 100 worsens overfitting by allowing more iterations beyond the divergence point. Increasing the learning rate causes training instability and does not address the generalization gap. Adding more layers increases model complexity and capacity to memorize, making overfitting worse. Other complementary techniques include dropout, L1/L2 regularization, and data augmentation.

Sourceaws.amazon.com

Create a free account

Question pool

399 practice questions

The Pass-IT question pool gives you material to practice for MLA-C01. A Pass-IT mock exam uses 65 questions and a 130-minute time limit; these are practice settings.

Pool details: MLA-C01

Passing score720 / 1,000

as published by AWS

Objectives in the guide12 objectives listed in the official guide

across 4 domains in the official exam guide

Pool size399 questions

= The pool size is equivalent to 6 sets of 65 questions; this does not mean that each mock exam uses a separate set.

Blueprint domains4 domains in the exam blueprint

Data Preparation for Machine Learning (ML) 95 · ML Model Development 104 · Deployment and Orchestration of ML Workflows 106 · ML Solution Monitoring, Maintenance, and Security 94

Recorded as checked against sources399 of 399

questions recorded as having their answer, options, and explanation checked against official AWS documentation

How questions are made →

Exam domains

What's on the exam

Data Preparation for Machine Learning carries the most weight at 28%, covering data ingestion and storage format choices, engineering new features and encoding raw values into something a model can consume, and the bias-detection and compliance work that has to happen before a model ever sees training data. ML Model Development follows at 26%, testing whether a candidate can pick the right way to build a model (an AWS AI service, a built-in SageMaker algorithm, or a foundation model), then train, tune, and evaluate it.

ML Solution Monitoring, Maintenance, and Security takes 24%, close behind model development itself — drift detection, infrastructure cost optimization, and IAM-level security for ML systems all live here. Deployment and Orchestration of ML Workflows closes the blueprint at 22%, covering the infrastructure and CI/CD choices that get a trained model into production; together the two domains put just under half the exam after training finishes.

Exam blueprint: MLA-C01

Data Preparation for Machine Learning (ML)28%

Ingest and store data from AWS sources using appropriate formats and services, and transform data through cleaning, feature engineering, and encoding techniques using tools such as SageMaker Data Wrangler. Also covers ensuring data integrity, addressing bias, and preparing data for model training in compliance with data protection requirements.

≈ 18 h

ML Model Development26%

Choose an appropriate modeling approach among AWS AI services, built-in algorithms, and foundation models, train and refine models through hyperparameter tuning and regularization, and analyze model performance using evaluation metrics and bias detection tools like SageMaker Clarify.

≈ 17 h

Deployment and Orchestration of ML Workflows22%

Choose deployment infrastructure suited to real-time, batch, or edge inference based on cost and latency, define that infrastructure as code with containers behind it, and automate the CI/CD pipelines that keep ML workflows shipping.

≈ 14 h

ML Solution Monitoring, Maintenance, and Security24%

Monitor model inference for drift and performance degradation using tools such as SageMaker Model Monitor, monitor and optimize ML infrastructure and costs, and secure AWS resources through IAM policies, network isolation, and auditing of ML systems.

≈ 16 h

Format

Exam format and question types

The exam draws 65 questions (50 scored, 15 unscored) from a mix of multiple-choice, multiple-response, ordering, and matching formats inside a 130-minute window. Questions center on SageMaker workflows: training job configuration, hyperparameter tuning, deployment target selection, and pipeline automation, framed as scenarios rather than definitions. At roughly 2 minutes per question, the longer scenarios deserve more of the clock than an even split would give them.

Question types: MLA-C01

Multiple Choice50%

Select the single answer that best meets the question’s requirements.

Multiple Response25%

Select multiple answers. Follow the question’s instructions on how many to choose.

Ordering15%

Arrange the steps in the sequence needed to complete the process.

Matching10%

Match each item with its corresponding item, following the question’s instructions.

See AWS for official question-format information. The shares shown describe the Pass-IT practice pool; they do not establish the proportions on the official exam.

Planning

Preparing for MLA-C01

The exam runs through Pearson VUE, either at a testing center or online with a remote proctor, and is offered in English, Japanese, Korean, and Simplified Chinese. Certification holds for 3 years; renewal means passing a recertification exam before it expires, or earning a higher-level AWS certification, which renews it automatically.

Preparation and logistics: MLA-C01

Preparation

Illustrative study time40–100 h

illustrative planning range: 40 h with relevant experience to 100 h when starting out; your needs may fall outside this range

LevelIntermediate

Taking and maintaining the certification

DeliveryPearson VUE testing center or online proctored exam

Retake policy14-day waiting period between exam attempts, no limit on total number of attempts

Certification validity3 years

Pass a recertification exam before the 3-year expiration date, or earn a higher-level AWS certification to automatically renew

Official guideView on provider site

Study notes

Common pitfalls

Topics to review: MLA-C01

01SageMaker Modes
Confusing SageMaker training jobs, processing jobs, real-time endpoints, batch transform, and serverless inference and when to use each deployment mode.
02Feature Engineering
Not understanding SageMaker Feature Store, data wrangling, and preprocessing pipeline design leads to wrong answers on data preparation questions.
03Model Monitoring
Misunderstanding SageMaker Model Monitor capabilities for detecting data drift, model drift, bias, and feature attribution changes in production.
04Pipeline Automation
Not knowing how to build end-to-end ML pipelines using SageMaker Pipelines, Step Functions, and EventBridge for automated retraining workflows.
05Cost Optimization
Choosing expensive real-time endpoints when batch transform or serverless inference would meet the latency and throughput requirements at lower cost.

Practice with questions and explanations in Pass-IT →

Questions

Frequently asked questions

How long is the AWS Certified Machine Learning Engineer - Associate exam?

The MLA-C01 exam has 65 questions and a 130-minute time limit.

What is the passing score for AWS Certified Machine Learning Engineer - Associate?

The passing score for the MLA-C01 exam is 720 / 1000.

Which pitfalls should I review when preparing for AWS Certified Machine Learning Engineer - Associate?

Topics to review include SageMaker Modes, Feature Engineering, Model Monitoring, Pipeline Automation, Cost Optimization. Work through examples to check that you understand the distinctions and can explain your answer.

Which MLA-C01 domains carry the most marks?

Data preparation is the largest at 28%, followed by model development at 26%, monitoring, maintenance and security at 24% and deployment and orchestration at 22%. That is close to an even split across four domains. The practical read is that MLA-C01 tests the whole lifecycle rather than modelling alone.

Should you take AIF-C01 before MLA-C01?

Not necessarily. AIF-C01 is a foundational exam about AI concepts and responsible use, while MLA-C01 is an associate engineering exam about building and operating pipelines on SageMaker. If you already train and deploy models, going straight to MLA-C01 saves you a sitting.

How long does MLA-C01 stay valid?

Three years from the pass date. Renew by sitting the exam again before it expires, or by earning a higher-level AWS certification, which renews it automatically.

What is the MLA-C01 retake policy?

AWS requires 14 days between attempts and sets no limit on how many you take. The waiting period applies from the first failure onwards, not only to later ones.

What do you need to know before booking MLA-C01?

There is no formal prerequisite. The catalog budget is roughly 65 hours, and the exam assumes you can already write Python against SageMaker and reason about data quality. It is an engineering exam, so preparation that stays theoretical tends to run short.

One certification, 12 months

Practice for MLA-C01

Focus your practice on one certification, or choose Pro to practice across all certifications.

Buy MLA-C01 access for $29.99

Start a free practice sessionTry the first 20 questions without a card to see whether the practice suits you.

For eligible purchases: money-back guarantee if you fail your exam.

View guarantee terms →