AWS opened beta registration for a rewritten version of the Machine Learning Engineer Associate certification exam on September 1, 2026, and it is not a minor content refresh. The certification that launched barely two years ago is getting rebuilt around generative AI, agentic workflows, and Amazon Bedrock, which tells you something about how fast the actual job has moved underneath the credential. If you already hold MLA-C01, are studying for it right now, or are trying to decide whether it is even worth pursuing this year, the next few weeks matter more than they normally would for a routine exam update.
Here is what AWS actually changed, why the timing is not an accident, and how to decide whether to sit the current exam before it closes or wait for the version that replaces it.
What’s Actually Changing Between MLA-C01 and MLA-C02
AWS did not touch the exam’s skeleton. The four domains that defined MLA-C01 carry over unchanged into MLA-C02: data preparation for machine learning, ML model development, deployment and orchestration of ML workflows, and ML solution monitoring, maintenance, and security. What changed is what AWS expects a certified engineer to actually know inside each one.
The Four Domains Stay the Same, the Content Inside Them Doesn’t
According to AWS’s own announcement, the task statements underneath those four domains were rewritten to reflect how the ML engineer role has broadened in practice. Traditional model-building and pipeline skills are still tested, but they now sit alongside content that did not exist when MLA-C01 launched.
The New Material: Generative AI, Agentic Workflows, and Amazon Bedrock
The updated exam adds real weight to:
- Fine-tuning and operationalizing foundation models rather than just consuming them
- Retrieval-augmented generation (RAG) architectures and when to use them over fine-tuning
- Orchestrating AI agents and multi-step agentic workflows in production
- Expanded Amazon Bedrock coverage, beyond the surface-level treatment it got in MLA-C01
- Responsible AI practices applied across both traditional ML and generative AI systems
That is a meaningfully different exam, even with the same four-domain outline. A candidate who only prepared with MLA-C01-era material will walk in underprepared for a noticeable slice of MLA-C02’s questions.
The Timeline Candidates Actually Need to Track
The dates here are unusually tight for an AWS certification transition, and missing one of them changes which exam you end up sitting.
- September 1, 2026: Beta registration for MLA-C02 opens in English, and the full exam guide with detailed task statements is published
- September 28, 2026: Last day to sit MLA-C01 in English
- September 29, 2026: Beta delivery of MLA-C02 begins
- Early 2027: MLA-C02 moves out of beta into its standard, generally available form
One detail that trips people up: MLA-C01 does not disappear everywhere on September 28. AWS is keeping it available in Japanese, Korean, and Simplified Chinese during the beta window, so the hard cutoff only applies to English-language testing.
Should You Sit MLA-C01 Now or Wait for MLA-C02?
There is no universally correct answer here, but the decision comes down to how far along your prep already is and how comfortable you are testing beta content.
Reasons to Test Before September 28
- You have already been studying MLA-C01 material and are close to exam-ready
- You want a stable, non-beta exam with a known question style and no untested items
- You need the credential fast for a role or contract requirement and cannot wait for early 2027
Reasons to Wait for the Beta
- Your work already involves Bedrock, RAG, or agentic AI, so the new content plays to your strengths rather than against them
- You want the credential that reflects current hiring language, since job postings increasingly reference generative AI and agent orchestration explicitly, not just “machine learning”
- You are comfortable with beta-exam mechanics, including the longer post-exam scoring wait that comes with any AWS beta
Beta exams also run cheaper than the standard registration fee, at $75 for the 170-minute, 85-question MLA-C02 beta versus the usual $150 AWS charges for a standard associate-level exam. That is a real incentive, not just a footnote, if your preparation already leans toward the new content.
Why AWS Is Rewriting a Two-Year-Old Exam Already
MLA-C01 is not old by certification standards, which makes the speed of this update notable on its own. AWS’s own announcement of the update states plainly that the ML engineer role itself has changed faster than the exam did. That tracks with what is happening across the certification industry generally this year: AWS already restructured its entry-level AI credentials once in 2026 after retiring the older Machine Learning Specialty certification, and other vendors are moving just as fast. Anthropic launched its own Claude Certification Program this year, and traditional cloud vendors are racing to keep their credentials matched to what generative AI and agentic systems actually require day to day in production.
The underlying labor trend supports the urgency. The U.S. Bureau of Labor Statistics projects 35% employment growth for data scientists between 2025 and 2035, ranking the occupation third among all fastest-growing jobs this decade, with median pay already at $120,230 as of 2025. AWS’s machine learning engineering role sits squarely inside that growth curve, and a meaningful share of the demand driving it now assumes generative AI fluency rather than treating it as a specialty add-on.
What This Certification Is Actually Worth Right Now
Salary and Demand Data
Reported salary figures tied to the AWS Machine Learning Engineer Associate credential range from roughly $105,000 to $175,000, with an average near $135,000 and a reported year-over-year increase around 12%. Broader AWS ML engineer pay data (not limited to certificate holders specifically) puts the U.S. average closer to $128,769 as of mid-2026, with location driving significant swings, San Francisco-based roles average well above $230,000. None of this is a guarantee tied to passing an exam. It reflects what the underlying job market is currently paying people doing this work, certified or not, and a certification is one input into how competitive a candidate looks for it.
What It Costs to Get and Keep This Credential
- Standard MLA-C01 exam fee: $150 (available through September 28, 2026, in English)
- MLA-C02 beta exam fee: $75, through September 28 to 29 registration window
- Recertification fee after the standard three-year cycle: $75, using AWS’s shorter recertification-exam format
- 180-day grace period after expiry during which recertification is still available at the discounted rate
Who Should Actually Pursue This Certification, and Who Should Skip It
- Pursue it if: you already work with SageMaker, Bedrock, or production ML pipelines and want a credential that now genuinely reflects that work
- Pursue it if: you are targeting roles that explicitly mention generative AI, LLM fine-tuning, or agentic workflows alongside traditional ML engineering duties
- Skip it for now if: you have zero hands-on AWS ML experience yet, since this is not an entry-level credential and the new content raises the bar rather than lowering it
- Skip it for now if: you specifically need broad AI literacy for a non-engineering role, where AWS Certified AI Practitioner is the better-fit, lower-cost starting point
How Machine Learning Engineer Associate Fits Into AWS’s AI Certification Lineup
AWS now has a genuine ladder of AI-focused credentials rather than one catch-all option, and it is easy to pick the wrong rung if you have not compared them directly.
| Certification | Level | Best For |
|---|---|---|
| AWS Certified AI Practitioner (AIF-C01) | Foundational | Non-engineers who need broad AI/ML literacy for business or product roles |
| AWS Certified Machine Learning Engineer – Associate (MLA-C01/C02) | Associate | Engineers who build, deploy, and operate ML and generative AI systems on AWS |
| AWS Certified Generative AI Developer – Professional (AIP-C01) | Professional | Experienced developers building generative AI applications specifically, not general ML pipelines |
Someone deciding between these three should think in terms of role, not just difficulty. A data engineer moving into ML operations belongs in the Associate tier; a solutions architect who needs to speak AI credibly in client conversations is usually better served starting at the Foundational tier instead.
DirectCertify is an independent certification prep provider and is not affiliated with, endorsed by, or sponsored by Amazon Web Services. Exam dates, pricing, and beta-registration details change on AWS’s own schedule, so confirm current specifics directly on AWS’s certification pages before registering.
