How to Pass the AWS ML Engineer Associate (MLA-C01/C02): A 6-Week Plan

About 65 hours over six weeks, weighted to the four exam domains, which keep the same structure in MLA-C01 and MLA-C02. AWS lists MLA-C01 at 150 USD, 65 questions in 130 minutes, with a passing score of 720 on a 100 to 1,000 scale (50 questions are scored and 15 are unscored). The MLA-C02 beta, delivered from September 29, 2026, is 75 USD, 85 questions and 170 minutes.

Read the next section before you book anything, because the date you sit this exam decides which version you get.

Before week one: pick your exam version

AWS has set September 28, 2026 as the last day for the English version of MLA-C01. The successor, MLA-C02, opened beta registration on September 1, 2026 and starts beta delivery September 29, 2026. The beta is 75 USD, 85 questions, 170 minutes, English only, and you can take it once. Japanese, Korean and Simplified Chinese versions of MLA-C01 stay available until the standard MLA-C02 launches, which AWS’s certification blog dates to January 14, 2027, though the exam page still lists general availability as TBD. On that date MLA-C01 retires in every language. On the AWS exam page the beta is listed under its own exam code, ME1-C02.

	MLA-C01	MLA-C02 beta
Price	150 USD	75 USD
Questions	65 (50 scored)	85
Time	130 minutes	170 minutes
Languages	English through Sep 28, 2026; JA, KO, ZH-CN until Jan 14, 2027	English only
Retakes	14-day wait, no attempt limit	One attempt; retake on the standard MLA-C02

Six weeks from any start date after mid-August 2026 lands after the English C01 cutoff. Your realistic options are the C02 beta while it runs, MLA-C01 in one of the three other languages until January 14, 2027, or the standard MLA-C02 from that date. The MLA-C02 exam guide keeps the same four domains, reweights them to 28/24/24/24, and adds foundation model, RAG and agent tasks alongside Amazon Bedrock. If you are aiming at C02, treat that material as a fifth topic and spread it across weeks 2 through 5. Our MLA-C02 exam guide lists what changed.

Then read the exam guide for your version and know the weights, because the hours below follow them.

Domain (C01 name / C02 name)	C01 weight	C02 weight	Plan hours
Data Preparation for ML / Data Preparation for ML and AI	28%	28%	16
ML Model Development / ML Model and Foundation Model (FM) Development	26%	24%	15
Deployment and Orchestration of ML Workflows / Deployment and Orchestration of ML and AI Workflows	22%	24%	13
ML Solution Monitoring, Maintenance, and Security / Operating, Monitoring, and Securing ML and AI Solutions	24%	24%	14

That is 58 hours of study. Week 6 adds about 7 hours of practice and review. On C02, move an hour from model development to deployment to match the new weights.

Weeks 1 and 2: data preparation (about 16 hours)

The biggest domain, and the one that tests the most services. Ingestion, transformation, feature engineering, and data quality, each with an AWS service attached.

Follow the Exam Prep Plan on AWS Skill Builder. It is the resource AWS points to from the exam page.
Do the work in an account. Land a dataset in S3, transform it with SageMaker Data Wrangler and a Glue job, store features in SageMaker Feature Store, and run a data quality check. Note the cost and the IAM you needed for each step.
Know formats and when they matter: Parquet versus CSV, when to use Kinesis versus batch, and how to handle imbalanced classes and missing data.

Week 3: model development (about 15 hours)

Algorithm selection, training, tuning, and evaluation. The exam expects you to know which SageMaker built-in algorithm fits a described problem, how to run hyperparameter tuning, and how to read a confusion matrix, precision, recall, F1, and RMSE.

Train at least two models yourself. One built-in algorithm, one custom script in a SageMaker training job. Set up a tuning job with a defined objective metric. This is the week reading gets you the least; AWS says the target candidate has a year of SageMaker experience, and the questions assume it.

Week 4: deployment and orchestration (about 13 hours)

Deploy the same model three ways: real-time endpoint, serverless inference, and batch transform. Know the cost and latency trade-off for each.
Build one SageMaker Pipeline end to end, then look at how Step Functions and EventBridge would trigger it.
Know your infrastructure-as-code options and CI/CD for ML at the level of which service does what.

Week 5: monitoring, maintenance and security (about 14 hours)

This domain is 24% of scored content on both versions, and it is the one experienced ML people underestimate. Set up SageMaker Model Monitor for data drift and model quality, know what CloudWatch and CloudTrail give you, and then spend a full evening on security: IAM roles for SageMaker, VPC configuration, KMS encryption, and least privilege for a training job that reads from S3. Cost optimization questions land here too, so know Spot training and right-sizing instances.

Week 6: practice and weak domains (about 7 hours)

Take the AWS Certification Official Practice Question Set and Official Pretest on Skill Builder, timed, and check which exam version each one covers before you rely on it. Then score yourself by domain and put the remaining hours into whichever is lowest. With every domain between 22% and 28% on either version, there is nowhere to hide a weak one.

Exam day

Pearson VUE, test center or online proctored. Online means starting check-in 30 minutes before your appointment, a government photo ID that exactly matches the name on your booking, a 360-degree room scan, a single monitor, and a cleared desk. Formats are multiple choice, multiple response, ordering and matching on MLA-C01; the MLA-C02 guide lists only multiple choice and multiple response. Both give you 2 minutes per question: 65 in 130 minutes on C01, 85 in 170 minutes on the beta. If English is not your first language, request AWS’s 30-minute extension before you book.

If you fail MLA-C01, AWS makes you wait 14 calendar days and charges the full fee again, with no limit on attempts. Beta exams can be taken only once; fail the MLA-C02 beta and your retake is the standard MLA-C02. Beta results typically arrive within 5 business days. The certification is valid for 3 years.

The rule that separates pass from fail

Many MLA questions describe a working system and ask for the change with the least operational overhead, or the lowest cost, or the fastest path. Read for that qualifier first. Two of the four options will usually work; only one meets the constraint. Six weeks of building in a real account is what makes that distinction feel obvious on the day.

Drill it week by week

Free MLA-C01 practice questions, weighted to the official blueprint, with a plain-English explanation on every one. The four domains carry over to MLA-C02.

Free practice questions →

Keep reading: MLA-C01 study guide · MLA exam cost · MLS-C01 vs MLA-C01 · All AI & cloud exam prep

HOW TO // AI is not affiliated with or endorsed by AWS. AWS Certified Machine Learning Engineer – Associate, MLA-C01, MLA-C02 and AWS are trademarks of Amazon Web Services, Inc. or its affiliates; we reference them descriptively. All content is original. Exam details come from AWS’s official pages as of September 2026 and can change; check the official exam page before booking.

Practice for every AI & cloud cert

Related guides

Frequently asked questions

How long does it take to study for the AWS Machine Learning Engineer Associate?

This plan uses about 65 hours over six weeks. AWS pitches the exam at people with at least 1 year of Amazon SageMaker experience plus 1 year in a related role, so budget extra lab time if you have not built on AWS before.

Should I study for MLA-C01 or MLA-C02?

In English, MLA-C01 ends September 28, 2026, so anyone sitting later studies for MLA-C02: the beta from September 29, 2026, or the standard version AWS has scheduled for January 14, 2027. MLA-C01 stays available in Japanese, Korean and Simplified Chinese until that date.

What is different about MLA-C02?

It keeps the same four domains, reweighted to 28/24/24/24, and adds tasks on foundation models, RAG, agents, Amazon Bedrock and responsible AI safeguards. A few MLA-C01 items, such as SageMaker Neo edge optimization and bring-your-own-container, were removed. The C02 guide lists only multiple choice and multiple response questions.

What happens if I fail?

For standard exams AWS makes you wait 14 calendar days and charges the full fee again, with no limit on attempts. The MLA-C02 beta can be taken only once; if you fail it, your retake is the standard MLA-C02.