AWS Certified Machine Learning Engineer Associate
275 practice questions
Last reviewed: April 2026
Personal notes and resource links for your study journey
Filter by Certification
The AWS Certified Machine Learning Engineer Associate (MLA-C01) launched in August 2024 as the practitioner-focused counterpart to the older Machine Learning Specialty. It validates the ability to build, deploy, monitor, and maintain ML workloads on AWS — with a strong emphasis on Amazon SageMaker, MLOps tooling, and production model lifecycles. The exam targets early- and mid-career ML engineers, data scientists moving into engineering, and DevOps engineers expanding into ML platforms. Expect scenario-driven questions about feature pipelines, model registries, deployment patterns, drift detection, and cost-aware inference. The exam is conceptual and hands-off (no labs), but assumes the candidate has actually shipped models to production.
The largest domain at 28%. SageMaker Data Wrangler, Feature Store, Glue, and S3 data lake patterns. Expect questions on handling imbalanced data, leakage, encoding strategies, and feature engineering at scale.
SageMaker training jobs, built-in algorithms vs. bring-your-own-container, JumpStart, hyperparameter tuning, and model evaluation. Common stumbling block: choosing between SageMaker Autopilot, Canvas, and custom training.
SageMaker endpoints (real-time, async, serverless, batch transform), Pipelines, Model Registry, and CI/CD with CodePipeline. Candidates often miss subtle differences between deployment modes and their cost tradeoffs.
SageMaker Model Monitor, Clarify (bias and explainability), drift detection, and IAM/VPC patterns for ML workloads. Tests practical MLOps fluency more than theory.
Services you'll encounter on the exam and why each one matters.
End-to-end ML platform covering notebooks, training jobs, hyperparameter tuning, processing jobs, managed inference endpoints, and MLOps pipelines.
Why it's on the exam: SageMaker is the umbrella service spanning all four MLA-C01 domains — expect questions on training infrastructure choices, inference deployment options, and managed-vs-self-hosted tradeoffs.
Web-based IDE for ML — Jupyter notebooks, experiments, Pipelines, Model Registry, JumpStart, and Canvas all in a single workspace.
Why it's on the exam: Domain 2 (ML Model Development) tests Studio as the unified surface for iterating on models, debugging training, and promoting artifacts.
Catalogue of pre-trained foundation and task-specific models with one-click deployment, transfer-learning notebooks, and fine-tuning workflows.
Why it's on the exam: JumpStart is the canonical answer when a question asks how to start from a pre-built model rather than train from scratch — relevant under Domain 2.
Bias-detection and explainability tool that produces SHAP feature attributions plus pre-training and post-training bias metrics on tabular and foundation models.
Why it's on the exam: Domain 2 + Domain 4 questions on responsible AI, model explainability, and fairness audits name Clarify as the AWS-native answer.
Continuously checks deployed endpoints for data-quality drift, model-quality drift, bias drift, and feature-attribution drift against a baseline.
Why it's on the exam: Domain 4 (Monitoring, Maintenance, and Security) repeatedly tests how to detect and respond to drift in production — Model Monitor is the named service.
Managed repository for ML features with synchronized online (low-latency) and offline (batch) stores, point-in-time correctness, and feature reuse across models.
Why it's on the exam: Domain 1 (Data Preparation) tests Feature Store as the canonical way to avoid training/serving skew and share features across teams.
Native MLOps orchestrator for SageMaker — chains preprocessing, training, evaluation, model registration, and conditional deployment as a versioned DAG.
Why it's on the exam: Domain 3 (Deployment & Orchestration) emphasizes reproducible end-to-end pipelines; Pipelines is the AWS-native choice over generic step orchestrators for SageMaker-heavy stacks.
Visual data-prep tool inside Studio for importing data from S3, Athena, Redshift, Snowflake, and applying 300+ built-in transformations with one click.
Why it's on the exam: Domain 1 questions on feature engineering and exploratory data analysis frequently name Data Wrangler as the low-code answer for tabular prep.
Object storage that serves as the data lake for training datasets, model artifacts, inference inputs/outputs, and SageMaker Feature Store offline data.
Why it's on the exam: Every MLA-C01 data-prep and model-deployment scenario assumes S3 as the data substrate; storage classes, lifecycle policies, and access patterns surface in Domains 1 and 4.
Serverless ETL service with a managed Spark runtime, a Data Catalog, crawlers for schema discovery, and Glue DataBrew for low-code transformation.
Why it's on the exam: Domain 1 names Glue as the default ETL/data-catalog tool for moving raw data into the shape SageMaker training expects.
Serverless interactive SQL engine over S3 (and federated sources), using the Glue Data Catalog for schema and pay-per-query pricing.
Why it's on the exam: Athena is the expected answer when a question asks how to run ad-hoc SQL on S3 training data without spinning up a cluster — common in Domain 1.
Managed Hadoop/Spark platform for large-scale data processing, supporting Spark MLlib, Hive, Presto, and SageMaker Studio EMR integration.
Why it's on the exam: EMR appears in Domain 1 scenarios that exceed Glue's scale or require Spark MLlib pipelines outside SageMaker.
Real-time data-streaming service for ingesting clickstream, IoT, and log events at scale, replayable within the retention window.
Why it's on the exam: Domain 1 questions on streaming feature ingestion (e.g. fraud detection, recommendation freshness) name Kinesis as the AWS-native answer.
Serverless compute for event-driven inference, lightweight preprocessing, S3 event triggers, and stitching SageMaker calls into business workflows.
Why it's on the exam: Domain 3 deployment scenarios distinguish "host on SageMaker endpoint" from "wrap in Lambda" tradeoffs — cost, cold start, and payload-size questions are common.
Managed container registry for the Docker images that SageMaker training jobs, processing jobs, and inference endpoints pull at runtime.
Why it's on the exam: Domain 3 tests bring-your-own-container (BYOC) workflows for custom training/inference — ECR is the named storage and IAM-integrated registry.
Serverless workflow orchestrator with native SageMaker integrations for training, batch transform, endpoint deployment, and Lambda step composition.
Why it's on the exam: Distinguishing Step Functions (multi-service orchestration) from SageMaker Pipelines (SageMaker-native MLOps) is a recurring Domain 3 distractor pattern.
Account-wide access control: users, roles, policies, federation, and least-privilege permissions for every SageMaker, S3, and pipeline action.
Why it's on the exam: Domain 4 (Security) tests IAM execution roles for training/inference, cross-account model sharing, and resource-based policies on the data lake.
Managed creation and control of cryptographic keys used to encrypt training data, model artifacts, EBS volumes on training instances, and endpoint payloads.
Why it's on the exam: Encryption-at-rest with customer-managed keys is the canonical Domain 4 answer for protecting sensitive training corpora and model IP.
Metrics, logs, and alarms for SageMaker endpoint invocations, training-job progress, custom model metrics, and pipeline step durations.
Why it's on the exam: Domain 4 expects CloudWatch for endpoint latency/error alarms, training-job log troubleshooting, and surfacing Model Monitor findings to ops teams.
Account-wide audit log of every API call — who launched a training job, who updated an endpoint, who downloaded model artifacts from S3.
Why it's on the exam: Compliance scenarios in Domain 4 cite CloudTrail as the immutable record needed to answer "who deployed this model" and "when was the training data accessed."
$120k–$165k–$230k USD annual
Range covers US-based mid-to-senior MLE roles where AWS proficiency is required. FAANG / unicorn senior MLEs frequently exceed $300k TC. Entry-level and non-coastal markets trend lower. The cert alone does not move salary — it complements a portfolio of shipped ML systems.
Source: levels.fyi 2025–2026 ML engineer roles, U.S. BLS OEWS May 2024 (15-2051 data scientists, 15-1252 software developers). Figures are approximate; actual compensation depends on role, region, and experience.
Demand for ML engineers who can productionize models — not just train them in notebooks — accelerated through 2024–2026 as enterprises operationalized GenAI and classical ML workloads. MLA-C01 functions as a credible signal that a candidate understands SageMaker end-to-end and can navigate MLOps tradeoffs. Recruiters at AWS-centric shops (financial services, healthcare, retail data teams) use it as a screening filter alongside Python and PyTorch/TensorFlow experience. It pairs naturally with the AI Practitioner (AIF-C01) and the Data Engineer Associate (DEA-C01) for a broader data-and-ML profile. It does NOT by itself qualify candidates for ML research roles, deep-learning specialist positions, or ML platform architect titles — those expect multi-year shipped-system experience plus often a graduate degree.
There are no formal prerequisites. AWS recommends at least one year of hands-on experience with SageMaker and ML workflows, plus working knowledge of Python, common ML libraries (scikit-learn, pandas, PyTorch or TensorFlow), and basic statistics.
The most efficient path is to pass AIF-C01 first (foundational AI vocabulary), then build a small end-to-end SageMaker project — feature store, training job, model registry, real-time endpoint, monitor — before sitting MLA-C01. Candidates with a Cloud Practitioner (CLF-C02) or Solutions Architect Associate (SAA-C03) background find the AWS-services questions much easier. A pure data-science background without AWS exposure is the hardest starting point and typically requires 80+ hours of additional service-specific study.
MLA-C01 is rated Associate and is meaningfully harder than AIF-C01 because it assumes hands-on SageMaker fluency. Plan 80–120 hours over 8–12 weeks if you have prior ML experience but limited AWS exposure; 40–60 hours over 4–6 weeks if you already work daily on AWS ML pipelines. The exam is 65 scored questions in 170 minutes — multiple-choice and multiple-response, no labs.
The most common stumbling block is the breadth of SageMaker sub-services (Studio, Pipelines, Feature Store, Model Registry, Clarify, Model Monitor, JumpStart, Canvas, Autopilot, Ground Truth) — questions often hinge on picking the right tool for a constrained scenario. The second pitfall is deployment modes: knowing precisely when to use real-time vs. async vs. serverless vs. batch transform endpoints, and the cost and latency tradeoffs of each.
Initial general availability. Beta exam ran in mid-2024. Replaces the older Machine Learning Specialty (MLS-C01) for engineering-focused candidates. Current version as of April 2026.
MLA-C01 (AWS Certified Machine Learning Engineer Associate) is a a moderately difficult exam expecting practical hands-on experience plus solid understanding of best practices Associate-level exam. Most candidates need 80–150 hours of study spread over 6–12 weeks for associate-level exams. Most candidates who score consistently above the passing threshold on practice exams pass on their first attempt.
Most candidates need 80–150 hours of study spread over 6–12 weeks for associate-level exams. Time-to-pass varies widely by prior experience. Engineers with hands-on production experience in the underlying technology typically need less; candidates new to the platform should plan toward the upper end of that range.
MLA-C01 is a recognized credential in the AWS ecosystem and signals validated knowledge to employers, recruiters, and clients. Whether it is worth the time and fee for you depends on your role and goals — it tends to pay off most for cloud engineers, architects, and consultants who work with AWS day-to-day or want to move into roles that do.
The passing score for MLA-C01 is 720 / 1000. The exam contains 65 questions and lasts 2 hr 50 min.
The MLA-C01 exam fee is $150 USD. Fees are set by AWS and may vary by region; always confirm the current price on the official AWS certification page before booking.
AWS certifications are valid for 3 years. Recertify by passing the current version of the same exam, or by passing a higher-level exam in the same path before expiration.
Yes. You can take the exam online (proctored via the provider's secure browser, available 24/7 in most regions) or at an in-person Pearson VUE test center during business hours. Both formats use the same questions, time limit, and passing score.
CertLabPro provides 15 study modes across the practice question bank for MLA-C01. The exam-simulation mode mirrors the real exam: 65 questions in 2 hr 50 min, with the same passing threshold of 720 / 1000. Browse mode lets you read every Q&A statically.