Decide which exam you can actually take
AWS began English MLA-C02 beta delivery on September 29, 2026. The English MLA-C01 exam ended September 28. AWS still offers C01 in Japanese, Korean, and Simplified Chinese until C02 general availability. The credential has not been retired; this is an exam-version transition. On its credential page, AWS calls the new version MLA-C02 but displays ME1-C02 as the beta booking code. Confirm both the version and bookable code in your AWS Certification Account before buying an exam or class.
- You need an English exam soon and already build ML systems on AWS: Read the C02 guide, then consider the beta only if its current scope and longer format fit your preparation. The beta is not a shortcut to an assured pass or immediate result.
- You need Japanese, Korean, or Simplified Chinese: C01 remains the listed exam for those languages during beta. Use C01 objectives and confirm the actual appointment before buying version-specific materials.
- You can wait or are new to ML engineering: You can compare instruction for a small ML workflow without buying an exam first. Do not buy an old C01 course on the assumption that its labs cover C02's foundation-model, retrieval, and agentic-workflow additions.
AWS sources disagree on the C02 general-availability date: its September 1 announcement says January 14, 2027, while the current credential page still marks GA registration and delivery TBD. Treat January 14 as an announced plan, not a confirmed appointment. Check the booking system and AWS page before relying on it or the corresponding C01 retirement date.
Who is this for, and how difficult is it?
AWS's C02 target profile describes at least one year using SageMaker AI, Bedrock, and related services for ML engineering, plus at least one year in a related technical role. It expects both traditional ML and generative-AI experience. This is recommended readiness background, not a formal prior-certification requirement stated on the checked AWS pages. The credential does not, by itself, prove that someone can operate a production ML system.
An optional fictional work sample: a team needs to detect faulty products from invented sensor records. Sketch data validation, a model choice, evaluation, deployment, a failure response, drift monitoring and access controls. Explain whether a foundation model is useful and what remains to test in an authorized isolated lab. Use invented data and a diagram; no cloud account or spending is needed for this paper exercise. It is a Learnerside exercise, not an AWS exam item or a pass predictor. You can ask for instruction immediately.
What the current C02 guide covers
These are percentages of scored C02 content, not an estimate of how many questions a particular beta candidate will see. AWS's C01-to-C02 comparison shows that the domain structure remains four parts, but foundation models, retrieval-augmented generation, agents, data protection, and AI operations appear more explicitly in C02. Do not use C01's 26% model-development and 22% deployment weights for C02.
| C02 content domain | Scored weight | Work to compare with instruction |
|---|
| Data Preparation for ML and AI | 28% | Can you prepare varied data, validate quality, and protect sensitive inputs? |
| ML Model and Foundation Model Development | 24% | Can you choose, tune, and evaluate an approach against quality, latency, and cost needs? |
| Deployment and Orchestration of ML and AI Workflows | 24% | Can you deploy repeatably and explain a retrieval or agent workflow and its failure handling? |
| Operating, Monitoring, and Securing ML and AI Solutions | 24% | Can you observe model behavior and spending, restrict access, and respond to failures? |
Use the full official guide for detailed tasks and service scope. The guide describes 50 scored and 15 unscored questions and a 720 scaled-score threshold for the standard exam, but it expressly qualifies its pass/fail discussion for beta. Do not apply that 65-question format or a promised 720 beta pass mark to the 85-question beta.
Exam cost versus training cost
As checked September 29, AWS lists the English C02 beta at 170 minutes, 85 multiple-choice or multiple-response questions, and USD 75. It lists the remaining-language C01 at 130 minutes, 65 questions, and USD 150. Both are offered through Pearson VUE at a test center or online with proctoring. The C02 general-availability price is not verified here. These issuer list prices are exam fees, not class prices or guaranteed regional checkout totals; taxes, currency, retakes, lab usage, and training can add cost. AWS says beta results are typically available within five business days, not instantaneously.
You can compare the public C02 guide and the AWS exam-prep path. The credential page also points to official practice questions, a pretest, digital courses, labs, and a practice exam. Access and price of individual Skill Builder resources were not verified; some labs or AWS services may cost money even when a reading resource is free. Any later cloud practice needs separate permission, usage-cost controls and cleanup; it is optional.
Paid instruction is worth considering when you need hands-on feedback on data pipelines, deployment, monitoring, or team scheduling, not just a repeat of the guide. No dated C02 provider seat, course authorization, tuition, voucher, recording, refund, or beta-specific attempt term has been verified for this page. Ask any seller to name the legal seller and delivery organization, C02 versus C01 courseware revision, dates and time zone, learner-operated labs, AWS usage charges, access period, instructor feedback, exam voucher and retake terms, and what happens if the exam changes again. Compare the total per learner, not a teaser tuition figure.
Roles, pay, and renewal
AWS names ML engineer, MLOps engineer, LLMOps engineer, data engineer, backend developer, and data scientist as related roles. These are work contexts, not jobs granted by a certificate. Occupation-level wage data cannot isolate MLA holders or prove a certification salary premium; this page makes no salary promise.
AWS lists three-year credential validity for both versions. Its recertification table lists passing the latest Machine Learning Engineer Associate exam or Generative AI Developer Professional exam as three-year renewal paths for an active credential. A 50% exam discount may exist in an individual AWS Certification Account; do not deduct it from a quote unless you have it and its terms apply.
Next action
You can ask about ML instruction now. For version-specific exam preparation, choose your language and likely month, verify the actual code and availability, and compare the C02 guide or the still-applicable C01 guide for your language. If guided practice would close those gaps, describe your timing and format or draft an employer approval brief. A request does not authorize sharing your details with a provider; that requires separate consent naming the recipient.
Sources and limits
- AWS credential and exam overview - version and beta code, language transition, formats, list fees, validity, and preparation links; checked September 29, 2026.
- AWS September certification announcement - announced January 14, 2027 GA delivery, which conflicts with the credential page's TBD display; checked September 29, 2026.
- AWS MLA-C02 exam guide and C01/C02 comparison - target profile, domains and version differences; checked September 29, 2026.
- AWS recertification table - active-credential renewal routes; checked September 29, 2026.
Learnerside is independent of AWS. This page paraphrases issuer material; it does not verify a provider offer or promise an exam result, job, salary, or future exam date.