The short version
MLA-C02 is the updated version of AWS’s Machine Learning Engineer Associate exam. It tests whether you can build, deploy, and operate machine learning and generative AI systems on AWS, not whether you can invent new algorithms. As of October 18, 2026, AWS lists this version as a beta, so a few details are not final. This guide separates what AWS has published from what it has not.
What the exam is
The credential is called AWS Certified Machine Learning Engineer – Associate. It sits at the associate level and replaces the earlier MLA-C01 version. AWS says the exam covers working with both traditional ML models and foundation models (FMs), the large pretrained models behind generative AI.
Think of the job it describes as the person who takes a data scientist’s idea and makes it run reliably. You prepare data, train or customize a model, put it behind an endpoint, wire up a delivery pipeline, and then keep an eye on it. AWS lists those jobs in its exam guide: ingest and validate data, pick a modeling approach and tune it, choose deployment infrastructure and scaling, build CI/CD pipelines, build agentic workflows, monitor, and secure everything.
The guide also lists what is out of scope. You are not expected to design full end-to-end AI solutions, set ML strategy, or work deeply in two or more ML specialties. That is useful to know when you decide how far to dig into a topic.
Who it is for
AWS describes the target candidate as someone with at least one year of experience using Amazon SageMaker AI, Amazon Bedrock, and other AWS services for ML engineering. It also suggests at least one year in a related role such as backend developer, DevOps developer, data engineer, or data scientist, plus experience with both traditional ML and generative AI. The prerequisites page turns that into a self-check.
Format at a glance
| Item | What AWS says (as of October 18, 2026) |
|---|---|
| Exam | MLA-C02, Associate level |
| Status | Beta. AWS’s September 2026 blog gives January 14, 2027 as the date general-availability delivery begins; the AWS certification page still said TBD on October 27, 2026 |
| Beta registration code | AWS’s certification page lists the beta as ME1-C02 |
| Time | 170 minutes for the beta |
| Questions | 85 for the beta; the exam guide describes 50 scored and 15 unscored |
| Question types | Multiple choice and multiple response |
| Passing score | 720 on a scale of 100 to 1,000 (exam guide) |
| Delivery | Pearson VUE test center or online proctoring |
| Languages | English only during the beta |
| Validity | 3 years |
The numbers for beta and the exam guide do not match, which is normal for a beta. The exam format page explains which one to trust and what to confirm when you book.
The four domains and their weightings
AWS gives weightings as a share of scored content.
| Domain | Weighting |
|---|---|
| Data Preparation for ML and AI | 28% |
| ML Model and Foundation Model (FM) Development | 24% |
| Deployment and Orchestration of ML and AI Workflows | 24% |
| Operating, Monitoring, and Securing ML and AI Solutions | 24% |
Data Preparation for ML and AI
Getting data in and making it usable: storage and streaming ingestion, data formats, transformation and feature engineering, vector databases and embeddings, preparing documents for retrieval, data quality, bias, and class imbalance.
ML Model and Foundation Model Development
Choosing an approach and training it: picking models and Amazon Bedrock foundation models, SageMaker AI algorithms and tuning, overfitting, fine-tuning and prompt engineering, and evaluating results with metrics, experiments, and human review.
Deployment and Orchestration
Getting models into production: compute and endpoint choices, real-time versus batch inference, containers, auto scaling, knowledge bases and retrieval pipelines, agents, and CI/CD with model registries.
Operating, Monitoring, and Securing
Keeping it healthy: drift and performance monitoring, cost control, observability tools, IAM, network isolation, encryption, logging, and safeguards such as Amazon Bedrock Guardrails.
What changed from MLA-C01
AWS’s comparison page shows the same four domain themes as before, with weightings moving from 28/26/22/24 to 28/24/24/24. The big change is content: retrieval augmented generation (RAG), embeddings, vector stores, Bedrock features, and agents are now in scope. If you learned from older MLA-C01 material, the domains explained page points out the gaps.
Official source
For AWS’s own pages, see the MLA-C02 exam guide and the certification page. They are the final authority if anything here differs.
More guides for MLA-C02
- MLA-C02 Study Plan: 4-Week and 8-Week
- MLA-C02 Sample Questions With Answers
- Is the MLA-C02 Exam Hard?
- Is MLA-C02 Worth It? An Honest Look
- MLA-C02 Cheat Sheet: Key Facts to Review
- AIF-C01 vs MLA-C02: Which First?
Next step
Read the domains explained page to see where your time should go. When you want to check yourself, our MLA-C02 practice tests are original practice sets built around these four domains.
Frequently asked questions
Is MLA-C02 still in beta?
Yes, it was still listed as a beta on October 27, 2026. An AWS blog from September 2026 says general-availability delivery begins January 14, 2027, but the AWS certification page still said TBD. Check the AWS certification page before you book.
Does the certification expire?
AWS lists a validity of three years. Its recertification page names two ways to renew: pass the latest version of the exam, or pass the AWS Certified Generative AI Developer – Professional exam. You can also maintain the credential for one more year in AWS Skill Builder, which needs a paid Skill Builder subscription and a credential within 90 days of expiry. The renewal page has the details.
Can I still take MLA-C01?
In English, the last day was September 28, 2026. AWS says the Japanese, Korean, and Simplified Chinese versions stay available until MLA-C02 reaches general availability.