AWS Machine Learning Engineer Associate: Cost, Format, and the MLA-C02 Beta

The most important fact about this exam in late 2026 is that the version you can sit depends on your language. AWS retired MLA-C01 in English after September 28, 2026; English candidates now sit the MLA-C02 beta, a longer and cheaper exam, until the new version reaches general availability on January 14, 2027. Japanese, Korean, and Simplified Chinese candidates keep MLA-C01 until that same date. This page covers that transition along with the verified cost, format, and renewal math, because buying study materials for the wrong version is the most expensive mistake available here.

Who this cert is actually for

This is AWS’s hands-on machine learning operations credential: it certifies that you can build, deploy, and maintain ML workloads on AWS. The MLA-C02 version widens that to generative AI, with AWS recommending at least a year of experience using Amazon SageMaker AI, Amazon Bedrock, and the surrounding ML engineering services, plus a year in a related role such as backend developer, DevOps, data engineer, or data scientist. The exam serves working ML and MLOps engineers, data engineers moving toward model deployment, and cloud engineers whose teams are shipping models and GenAI features rather than just experimenting.

Skip it if you want a first AI credential: AWS AI Practitioner covers AI fluency at $100 with no experience assumption, and jumping straight to the engineer tier without SageMaker or Bedrock time behind you buys a hard exam and a hollow signal. Skip it too if your ML work lives on another platform; the exam is AWS-service-specific, and Databricks’ engineering credentials or vendor-neutral study serve a multi-cloud career better. Holding this credential auto-recertifies AI Practitioner, which is a real convenience for people who started there.

What it costs all-in

The standard exam is $150, AWS’s Associate-tier price, sat at a Pearson VUE testing center or online proctored. AWS also publishes local Associate-tier prices: €128, A$224, ¥20,000, and ₹12,829.50 (rupee vouchers through Pearson’s Mindhub store only). Until January 14, 2027, English candidates pay the $75 beta price instead, with a catch covered below: a beta can be sat only once.

There is no course requirement. AWS publishes the exam guide free and runs an Exam Prep Plan on Skill Builder; a third-party course plus a practice set built for MLA-C02 runs $30 to $50, and the version label matters more than the brand. A failed standard attempt costs another full $150 after a 14-day wait. Passing any AWS exam also deposits a 50 percent discount voucher in your certification account, which the discounts guide covers in detail.

The exam itself

MLA-C01, still offered in Japanese, Korean, and Simplified Chinese, runs 65 questions in 130 minutes. The MLA-C02 exam guide keeps the same scoring: 50 scored questions plus 15 unscored ones AWS does not identify, multiple-choice and multiple-response, reported on a 100-to-1,000 scale with 720 to pass. Its four domains are Data Preparation for ML and AI (28 percent), ML Model and Foundation Model Development (24 percent), Deployment and Orchestration of ML and AI Workflows (24 percent), and Operating, Monitoring, and Securing ML and AI Solutions (24 percent). AWS says the domain structure is unchanged from MLA-C01 and that the new content adds generative AI implementation, retrieval-augmented generation, agentic AI workflows, foundation-model selection and fine-tuning, and expanded Amazon Bedrock coverage. AWS does not publish a pass rate for this or any exam.

The beta is a different sitting: 85 questions in 170 minutes for $75, English only, through January 14, 2027. AWS’s beta policy says beta candidates can take the beta once; fail it and you wait for the standard exam at general availability. Results for any AWS exam, beta included, post within five business days. If you already hold deep MLA-C01 knowledge and test in English, the old version is no longer an option; prepare against the MLA-C02 outline, and be skeptical of any course that does not name it. Stale-version prep is a recurring trap we document across vendors in the retirement playbook.

Renewal and the recertification voucher math

The credential is valid three years, and AWS’s recertification page lists three ways to keep it: pass the latest version of this exam, earn AWS Certified Generative AI Developer Professional (which recertifies this Associate credential), or maintain it on AWS Skill Builder, which adds one year at a time, opens within 90 days of expiry, and requires a paid Skill Builder subscription. The page also says the retake route can use the 50 percent voucher, so a holder who renews on the voucher pays $150, then $75, then $75, about $300 across nine years, before any employer reimbursement. The same voucher chain applies across AWS’s catalog, so this credential also discounts a move up to Solutions Architect Professional or sideways into the security specialty.

Nine-year cost of ownership: $300 to $450, the $150 exam plus two renewals at $75 each with the voucher or $150 each at full price. No annual fee applies. See certification renewal costs explained for how that compares.

What it does for the occupation you are entering

The credential maps to the data scientist occupation (BLS code 15-2051), the closest federal category to machine learning engineering, since BLS does not break ML engineers out on their own. The panel on this page shows the national median wage and headcount for that occupation as context for the field, not a wage this certificate produces. In hiring terms, the credential’s job is to corroborate hands-on claims: a resume that says “deployed models on SageMaker” plus this certification reads as verified rather than aspirational, which matters most for candidates crossing from adjacent roles (data engineering, cloud operations) into ML work.

Common mistakes

Sitting the beta as a practice run. At $75 the MLA-C02 beta looks like a cheap rehearsal, but AWS allows one beta attempt. Fail it and the next try is the full $150 standard exam, no earlier than January 14, 2027.

Studying MLA-C01 material for an English sitting. English MLA-C01 is gone. Courses that do not name MLA-C02, or that skip Bedrock, RAG, and agentic workflows, are teaching the old outline.

Skipping AI Practitioner when it is the better fit. If you cannot yet describe a SageMaker training job you have run, the associate engineer exam will spend your $150 teaching you that. The $100 AI Practitioner is the honest on-ramp, and this credential auto-recertifies it later.