All Certifications/MLA-C02

AssociateAdvanced

AWS Machine Learning Engineer Associate (MLA-C02)

Build, deploy, and operate ML and generative AI solutions on AWS

Updated for MLA-C02. MLA-C01 retired in English on September 28, 2026.

See the full timeline

Exam Quick Facts

Exam Code

MLA-C02

Questions

Duration

170 minutes

Passing Score

720/1000 (~72%)

Cost

$75 USD (beta pricing)

Valid For

3 years

Exam Version Since

2026

Difficulty

Advanced

Free cheat sheet14 pages · PDF

AWS Certified Machine Learning Engineer – Associate Cheat Sheet (MLA-C01)

Every in-scope SageMaker feature, MLOps workflow, monitoring and security decision the MLA-C01 exam actually tests — organised by the four exam domains with runbooks and decision tables

Certification Overview

The AWS Certified Machine Learning Engineer Associate (MLA-C02) validates your ability to build, operationalize, deploy, and maintain ML and AI solutions and pipelines on AWS. It replaced MLA-C01, which retired in English on September 28, 2026. MLA-C02 is in beta until general availability on January 14, 2027: the beta runs 85 questions in 170 minutes at $75 USD, English only. At general availability the exam guide lists 65 questions, 50 of them scored.

The big change is scope. MLA-C01 was about classic ML engineering on SageMaker. MLA-C02 keeps all of that and adds foundation models: choosing and customizing models in Amazon Bedrock, building RAG pipelines on vector stores, deploying agents, and monitoring the cost and quality of generative AI in production. AWS now expects candidates to have worked with both traditional ML and generative AI.

It is still an engineering exam rather than a data science one. The questions are about getting models — traditional or foundation — from notebook to production reliably, securely, and at a sensible cost.

Who Should Take This Exam?

ML engineers, MLOps engineers, data engineers, and backend developers building ML or generative AI systems on AWS should pursue this certification. AWS's target candidate has at least a year with SageMaker AI, Amazon Bedrock, and related services, plus a year in a role such as backend development, DevOps, data engineering, or data science.

It is also a strong next step after the AI Practitioner for cloud and infrastructure engineers moving into AI. The AI Practitioner teaches the vocabulary; this exam tests whether you can build and operate the systems.

Exam Domains

Data Preparation for ML and AI

28%

Still the largest domain. Ingesting, transforming, and validating data with S3, Glue, Glue DataBrew, Glue Data Quality, and SageMaker Data Wrangler and Feature Store. New for C02: vector databases (OpenSearch Service, RDS with pgvector), embedding models, chunking documents for RAG, masking and redacting sensitive data, and preparing data for FM fine-tuning and distillation.

ML Model and Foundation Model (FM) Development

24%

Choosing algorithms and modeling approaches, training and hyperparameter tuning, and evaluating models. New for C02: selecting foundation models in Amazon Bedrock, fine-tuning versus prompt engineering, RAG architecture patterns, performance/latency/cost trade-offs, NLP metrics such as BLEU, ROUGE, and BERTScore, human evaluation, and LLM-as-a-judge.

Deployment and Orchestration of ML and AI Workflows

24%

SageMaker AI endpoints (real-time, serverless, asynchronous, batch), auto scaling, SageMaker Pipelines, and CI/CD for ML. New for C02: FM deployment options, Bedrock Custom Model Import, Bedrock knowledge bases and retrieval pipelines, deploying agents and agent state, GPU scaling, and prompt management and prompt testing in pipelines.

Operating, Monitoring, and Securing ML and AI Solutions

24%

Model Monitor drift detection, CloudWatch, cost optimization, IAM execution roles, VPC isolation, and encryption. New for C02: monitoring agents and Bedrock evaluations, FM inference and token cost management, Bedrock Guardrails, Bedrock API keys versus IAM credentials, and scanning CI/CD pipelines with Amazon Inspector and CodeGuru.

Ready to test your MLA-C02 knowledge?

Practice with our free Machine Learning Engineer Associate (MLA-C02) exam — 65 questions with detailed explanations. No signup required.

Start Free Practice Exam

Study Strategy

Start from the classic ML engineering core, because it is still most of the exam: data preparation, SageMaker built-in algorithms, evaluation metrics, the four inference options, SageMaker Pipelines, and Model Monitor. Material written for MLA-C01 covers this well.

Then close the generative AI gap that C02 adds. Learn how Amazon Bedrock model selection, knowledge bases, Guardrails, and Custom Model Import work; how a RAG pipeline is built (chunking, embeddings, vector store, retrieval, reranking); how agents are deployed and monitored; and how token usage drives cost. AWS's official comparison of MLA-C01 and MLA-C02 lists every added and removed skill — use it as your checklist.

Hands-on practice helps more here than on most associate exams: train and deploy one model on SageMaker AI, and build one small RAG application on Bedrock. Delete endpoints when you are done. Budget 6-8 weeks with ML or data engineering experience, and longer if ML is new to you.

Key AWS Services to Know

Amazon SageMaker AI — training, endpoints, Pipelines, Feature Store, Model Monitor, Clarify
Amazon Bedrock — model selection, knowledge bases, Guardrails, Custom Model Import, Prompt Management
Amazon Bedrock AgentCore — deploying and operating agents
Amazon OpenSearch Service — vector search for RAG
AWS Glue, Glue DataBrew, and Glue Data Quality — ETL and data validation
Amazon S3 — data lake storage for ML datasets
AWS Step Functions — ML and AI workflow orchestration
Amazon CloudWatch — monitoring endpoints, models, and agents
AWS IAM and AWS KMS — execution roles, credentials, and encryption
Amazon Inspector and Amazon CodeGuru — securing ML CI/CD pipelines

Video: MLA-C02 Exam Preparation

Career Impact

ML engineering is one of the fastest-growing and highest-paying specializations in tech, and MLA-C02 now covers the generative AI work most teams are hiring for. It positions you at the intersection of software engineering, machine learning, and GenAI — a space where demand far exceeds supply. ML engineer roles on AWS typically pay $140K-$200K, with senior roles at top companies exceeding $250K.

This certification is particularly valuable as organizations move from AI experimentation to production. Teams that have prototypes in notebooks or chat playgrounds are hiring engineers who can deploy, monitor, and secure them — and this certification proves you can.

Frequently Asked Questions

What changed from MLA-C01 to MLA-C02?

The four domains keep the same shape, but MLA-C02 adds foundation models and generative AI throughout: Amazon Bedrock model selection and customization, RAG pipelines and vector databases, agents, generative AI evaluation metrics, and FM cost monitoring. Domain weights shifted to 28/24/24/24. A few C01 topics were removed, including SageMaker Neo edge optimization and bring-your-own-container.

Can I still take MLA-C01?

Not in English: the last day was September 28, 2026. MLA-C01 remains available in Japanese, Korean, and Simplified Chinese until MLA-C02 reaches general availability on January 14, 2027.

Is study material for MLA-C01 still useful?

Yes, for most of the exam. Data preparation, SageMaker algorithms, deployment, MLOps, and monitoring carry over. Add dedicated study of Amazon Bedrock, RAG, agents, and generative AI evaluation, which MLA-C01 material covers lightly or not at all.

What's the difference between the ML Engineer and the AI Practitioner?

The AI Practitioner is conceptual — no hands-on engineering. The ML Engineer Associate tests building and operating ML and generative AI systems: data pipelines, SageMaker AI training and endpoints, Bedrock and RAG, MLOps, monitoring, and security.