AWS Certified Machine Learning Engineer – Associate (MLA-C02) Exam Guide 2026

Important Update (2026): MLA-C01 is being replaced by MLA-C02. AWS is updating the AWS Certified Machine Learning Engineer – Associate certification to reflect the evolving role of modern ML engineers.

The updated MLA-C02 exam expands beyond traditional machine learning engineering to include Generative AI, Amazon Bedrock, Retrieval-Augmented Generation (RAG), Agentic AI, Foundation Models, Large Language Models (LLMs), and Responsible AI.

Registration for the updated exam MLA-C02 beta opened on September 1, 2026. During the beta phase, the exam is registered under the temporary beta code ME1-C02.

The last day to take the current MLA-C01 exam in English is September 28, 2026, while beta delivery of MLA-C02 begins on September 29, 2026.

If you already hold or pass the MLA-C01 certification before its retirement, your certification remains valid for its full 3-year validity period.

The MLA-C02 beta is currently available in English only. Additional languages, including Japanese, Korean, and Simplified Chinese, are planned for the general availability release.

If you are starting your preparation now, MLA-C02 is an excellent choice for validating both traditional ML engineering and modern GenAI/LLM skills on AWS.

By the end of this guide, you’ll have a clear understanding of what to expect from the AWS Certified Machine Learning Engineer – Associate MLA-C02 exam and how this certification can strengthen your career in machine learning, MLOps, LLMOps, and Generative AI.

What’s New in MLA-C02?

The updated MLA-C02 exam reflects the broadened scope of the modern ML engineer role.

The four-domain structure remains the same, but the existing task statements and skills have been updated to align with current industry practices.

Key additions include:

• Generative AI implementation – Building and deploying generative AI solutions using AWS services.
• Amazon Bedrock – Using Amazon Bedrock for foundation model access, customization, and GenAI application development.
• Retrieval-Augmented Generation (RAG) – Implementing architectures that combine foundation models with enterprise or external knowledge sources.
• Agentic AI – Orchestrating AI agents and building complex, multi-step AI workflows.
• Foundation Models and LLMs – Selecting, customizing, fine-tuning, and operationalizing foundation models and large language models.
• Responsible AI – Applying responsible AI practices across traditional ML and generative AI workloads.
• LLMOps – Applying operational practices to modern LLM and GenAI workloads.
• Modern ML engineering – Expanding traditional ML engineering practices to support increasingly complex AI workloads.

MLA-C01 vs MLA-C02: A Quick Comparison

Key Dates at a Glance

Important: During the Beta phase, the updated MLA-C02 exam is registered under the temporary exam code ME1-C02. Once the exam reaches General Availability, it will officially be available as MLA-C02.

What is the AWS Certified Machine Learning Engineer – Associate MLA-C02 Exam?

The AWS Certified Machine Learning Engineer – Associate (MLA-C02) is an associate-level certification offered by Amazon Web Services (AWS).

The certification validates a candidate’s ability to implement machine learning workloads in production and operationalize ML solutions using AWS services.

The updated MLA-C02 exam broadens this scope to include modern AI technologies such as:

• Generative AI
• Foundation Models
• Large Language Models (LLMs)
• Retrieval-Augmented Generation (RAG)
• Agentic AI
• Amazon Bedrock
• Responsible AI

The certification is designed for professionals who work with Amazon SageMaker AI, Amazon Bedrock, and other AWS machine learning and AI services.

Topics Included in MLA-C02 Exam

The AWS Certified Machine Learning Engineer – Associate MLA-C02 exam retains the four-domain structure of the current MLA-C01 certification.

However, the skills and task statements within those domains have been updated to reflect modern machine learning and AI engineering practices.

The four major domains continue to cover:

1. Data Preparation for Machine Learning (ML)
2. ML Model Development
3. Deployment and Orchestration of ML Workflows
4. ML Solution Monitoring, Maintenance, and Security

The updated exam introduces additional emphasis on Generative AI, Amazon Bedrock, RAG, Agentic AI, Foundation Models, LLMs, and Responsible AI across the existing domains.

Domain 1: Data Preparation for Machine Learning (ML)

Data preparation remains a critical component of the MLA-C02 exam.

Candidates should understand how to ingest, store, transform, prepare, validate, and secure data used in machine learning and AI workloads.

Key Topics

1. Ingest and store data:

• Structured and unstructured data
• Common data formats such as CSV, JSON, Parquet, ORC, Avro, and RecordIO
• Amazon S3 and other AWS storage services
• Streaming data sources
• Amazon Kinesis
• Apache Kafka
• Apache Flink
• Selecting appropriate AWS storage based on workload requirements

2. Transform data and perform feature engineering:

• Data cleaning
• Data transformation
• Outlier detection
• Deduplication
• Feature engineering
• Scaling and standardization
• Binning
• Encoding techniques
• Tokenization
• AWS Glue
• SageMaker Data Wrangler
• Data labeling and annotation
• SageMaker Ground Truth

3. Prepare data for ML and AI workloads:

• Data quality
• Data validation
• Data imbalance
• Bias identification
• Techniques for reducing data bias
• Data encryption
• Data anonymization
• PII and sensitive data protection
• Compliance requirements

4. Prepare data for Generative AI and LLM workloads:

• Preparing enterprise data for RAG architectures
• Data ingestion for knowledge-based AI applications
• Document processing
• Chunking and preprocessing concepts
• Creating high-quality data for foundation model applications
• Preparing information for retrieval-based GenAI applications

Domain 2: ML Model Development

MLA-C02 continues to test traditional machine learning model development while expanding into foundation models and Generative AI.

Key Topics

1. Choose a modeling approach:

• Selecting ML algorithms based on business requirements
• Supervised and unsupervised learning
• AWS AI services
• Amazon SageMaker built-in algorithms
• Model selection
• Model interpretability
• Selecting between traditional ML and foundation-model-based approaches
• Understanding when to use managed AI services

2. Train and refine models:

• Training processes
• Epochs
• Batch size
• Hyperparameter tuning
• Regularization
• Dropout
• Preventing overfitting and underfitting
• TensorFlow
• PyTorch
• SageMaker training capabilities
• Model experimentation
• Model version management

3. Analyze model performance:

• Accuracy
• Precision
• Recall
• F1 score
• RMSE
• ROC-AUC
• Model validation
• Overfitting and underfitting
• Model convergence
• SageMaker Clarify
• Model debugging and analysis

4. Generative AI, Foundation Models and LLMs

MLA-C02 introduces additional knowledge requirements around modern foundation-model workloads.

Candidates should understand:

• Foundation models
• Large Language Models (LLMs)
• Selecting an appropriate foundation model
• Model customization
• Fine-tuning concepts
• Prompt-based approaches
• Model evaluation
• Foundation model operationalization
• Generative AI application development
• Amazon Bedrock
• Responsible AI considerations for GenAI

5. Retrieval-Augmented Generation (RAG)

Candidates should also understand the purpose and architecture of RAG solutions, including:

• Why RAG is used
• Connecting foundation models to external knowledge
• Retrieval and generation workflows
• Enterprise knowledge integration
• Improving response relevance
• Reducing reliance on model-only knowledge
• Components involved in RAG architectures
• Operational considerations for RAG applications

Domain 3: Deployment and Orchestration of ML Workflows

MLA-C02 continues to emphasize the deployment and orchestration of production ML workflows.

The updated exam also expands this area to include deployment and orchestration considerations for modern AI and GenAI workloads.

Key Topics

1. Select deployment infrastructure based on architecture and requirements:

• Model deployment strategies
• Model versioning
• Rollback strategies
• SageMaker endpoints
• Real-time inference
• Batch inference
• Serverless inference concepts
• Kubernetes
• Amazon ECS
• Amazon EKS
• AWS Lambda
• Edge deployment
• Infrastructure requirements for ML workloads

2. Create and script infrastructure:

• Infrastructure as Code
• AWS CloudFormation
• AWS CDK
• Provisioning compute resources
• Training infrastructure
• Inference infrastructure
• Auto Scaling
• Capacity planning
• Cost considerations

3. Automate ML workflows:

• CI/CD
• AWS CodePipeline
• AWS CodeBuild
• AWS CodeDeploy
• SageMaker Pipelines
• Amazon EventBridge
• Automated training
• Automated model deployment
• Automated retraining
• Workflow orchestration

4. GenAI and Agentic AI Workflows

The updated MLA-C02 exam expands orchestration concepts to include:

• Generative AI application workflows
• Amazon Bedrock integrations
• AI agent orchestration
• Agentic AI workflows
• Multi-step AI workflows
• Tool usage by AI agents
• Coordinating complex AI tasks
• Operationalizing AI agent workloads
• Integrating AI agents with AWS services

Domain 4: ML Solution Monitoring, Maintenance, and Security

Production ML systems require continuous monitoring, maintenance, optimization, and security.

MLA-C02 continues to cover these traditional ML engineering responsibilities while extending them to GenAI and LLM workloads.

Key Topics

1. Monitor model inference:

• SageMaker Model Monitor
• Data drift
• Model drift
• Data quality monitoring
• Model performance monitoring
• Anomaly detection
• SageMaker Clarify
• Inference monitoring

2. Monitor and optimize infrastructure and costs:

• Amazon CloudWatch
• Infrastructure performance
• Resource utilization
• Latency
• Scalability
• Availability
• Troubleshooting
• Cost optimization
• AWS Cost Explorer
• AWS Trusted Advisor

3. Secure AWS ML resources:

• AWS IAM
• IAM roles
• IAM policies
• Access controls
• VPC configuration
• Encryption
• Secure ML pipelines
• Auditing
• Logging
• Compliance

4. Responsible AI

Responsible AI has an expanded role in MLA-C02.

Candidates should understand responsible AI considerations across both traditional ML and Generative AI workloads, including:

• Bias
• Fairness
• Transparency
• Explainability
• Privacy
• Security
• Responsible model usage
• Responsible GenAI implementation
• Monitoring AI system behavior
• Managing risks associated with AI-generated outputs

Amazon Bedrock in MLA-C02

One of the major additions to MLA-C02 is the expanded coverage of Amazon Bedrock.

Candidates should become familiar with how Amazon Bedrock can be used to build and operationalize Generative AI applications.

Important areas include:

• Foundation models
• Model selection
• Generative AI application development
• Model customization
• Fine-tuning concepts
• RAG architectures
• AI agents
• Agentic workflows
• Security
• Monitoring
• Responsible AI
• Production deployment considerations

Foundation Models and LLMs

MLA-C02 expands the certification’s scope into foundation models and large language models.

Candidates should understand how to:

• Select an appropriate foundation model
• Identify model capabilities and limitations
• Customize foundation models
• Understand fine-tuning
• Operationalize LLM workloads
• Evaluate model performance
• Integrate LLMs into applications
• Select appropriate approaches for GenAI use cases
• Consider security, cost, latency, and performance requirements

Agentic AI in MLA-C02

Another major addition is Agentic AI.

Modern AI applications can use agents to perform multiple steps, interact with tools, retrieve information, and execute complex workflows.

Candidates should understand:

• What AI agents are
• Agent orchestration
• Multi-step workflows
• Tool integration
• AI agent decision-making
• Workflow automation
• Agent-based application architectures
• Monitoring and operational considerations
• Security and responsible AI considerations for agents

AWS MLA-C02 Hands-on Preparation

Hands-on experience is an important part of preparing for the AWS Certified Machine Learning Engineer – Associate certification.

Candidates should practice working with AWS services used across the ML and GenAI lifecycle.

Recommended areas include:

• Amazon SageMaker AI
• Amazon Bedrock
• Amazon S3
• AWS Glue
• SageMaker Data Wrangler
• SageMaker Ground Truth
• SageMaker Pipelines
• SageMaker Model Monitor
• SageMaker Clarify
• Amazon CloudWatch
• AWS IAM
• AWS Lambda
• Amazon ECS
• Amazon EKS
• AWS CloudFormation
• AWS CDK
• Amazon EventBridge
• AWS CI/CD services

For MLA-C02 specifically, candidates should add practical exposure to:

• Generative AI applications
• Foundation models
• LLM workflows
• RAG architectures
• Amazon Bedrock
• AI agents
• Agentic workflows
• Responsible AI

Why Pursue AWS Certified Machine Learning Engineer – Associate MLA-C02?

The AWS Certified Machine Learning Engineer – Associate certification validates your ability to implement and operationalize machine learning workloads on AWS.

With MLA-C02, the certification also demonstrates exposure to the technologies shaping the next generation of AI engineering, including:

• Generative AI
• Foundation Models
• LLMs
• RAG
• Agentic AI
• Amazon Bedrock
• Responsible AI

This makes MLA-C02 particularly relevant for professionals looking to build careers across ML engineering, MLOps, LLMOps, data engineering, and AI application development.

The certification can help demonstrate practical knowledge of building, deploying, monitoring, and maintaining modern AI workloads on AWS.

Career Paths and Opportunities

MLA-C02 Beta Exam Information

The MLA-C02 Beta exam gives candidates an opportunity to take the updated certification before its general availability release.

During the Beta:

• The exam is available in English only.
• The Beta exam code is ME1-C02.
• The exam contains 85 questions.
• The exam duration is 170 minutes.
• The Beta price is $75 USD.
• Beta exams may contain additional questions used for statistical evaluation.
• These additional evaluation questions do not affect the candidate’s score.
• Beta results are typically available within 5 business days of exam completion.

Once the Beta phase is completed and the exam reaches General Availability, the certification will be officially available under the MLA-C02 exam code.

Exam Results for AWS Certified Machine Learning Engineer – Associate MLA-C02

AWS certification exams use a scaled scoring system.

Candidates should refer to the official MLA-C02 exam guide for the current scoring and exam-result details applicable to the updated examination.

The important point is that certification decisions are based on the overall exam result rather than requiring candidates to pass every individual domain separately.

AWS may also provide domain-level performance information to help candidates understand areas of relative strength and weakness.

MLA-C01 or MLA-C02: Which Exam Should You Take?

If you are already well prepared for MLA-C01, you can still take the current English exam before September 28, 2026.

Passing MLA-C01 before retirement does not shorten the certification’s validity. Your certification remains active for its full 3-year validity period.

However, candidates starting preparation now may prefer MLA-C02 because it reflects the broader modern ML engineering landscape.

Choose MLA-C01 if:

• You are already deeply prepared for MLA-C01.
• You want to take the current exam before it retires.
• You are comfortable with traditional ML engineering topics.
• You want to complete the certification before September 28, 2026.

Choose MLA-C02 if:

• You are starting your preparation now.
• You want exposure to Generative AI.
• You want to learn Amazon Bedrock.
• You want to understand RAG architectures.
• You want foundation model and LLM knowledge.
• You are interested in Agentic AI.
• You are targeting AI Engineer, MLOps, or LLMOps roles.
• You want a certification aligned with modern AI engineering practices.

Conclusion

The AWS Certified Machine Learning Engineer – Associate MLA-C02 represents the next evolution of AWS’s associate-level ML engineering certification.

While the core four-domain structure remains intact, the updated exam reflects how machine learning engineering has evolved from traditional model development into a broader discipline involving Generative AI, foundation models, LLMs, LLMOps, RAG, AI agents, automation, operationalization, and responsible AI.

For candidates already prepared for MLA-C01, taking the current exam before September 28, 2026 remains a valid option, and the certification retains its full validity period.

For candidates starting fresh, MLA-C02 provides an opportunity to validate a broader set of modern ML and AI engineering skills.

With technologies such as Amazon SageMaker AI, Amazon Bedrock, foundation models, RAG, and Agentic AI becoming increasingly important, the MLA-C02 certification can be a valuable addition to the profile of professionals targeting careers in ML Engineering, MLOps, LLMOps, Data Engineering, and AI/ML development.