{"id":26327,"date":"2026-10-06T07:55:28","date_gmt":"2026-10-06T07:55:28","guid":{"rendered":"https:\/\/www.examlabs.com\/certification\/?p=26327"},"modified":"2026-10-06T07:55:28","modified_gmt":"2026-10-06T07:55:28","slug":"amazon-mla-c01-pre-c02-exam-blueprint","status":"publish","type":"post","link":"https:\/\/www.examlabs.com\/certification\/amazon-mla-c01-pre-c02-exam-blueprint\/","title":{"rendered":"Amazon MLA-C01: Pre-C02 Exam Blueprint"},"content":{"rendered":"

MLA-C01 is now in transition rather than being the current English exam. AWS ended English delivery of MLA-C01 on September 28, 2026, and opened the English-only MLA-C02 beta on September 29. MLA-C01 continues in Japanese, Korean, and Simplified Chinese during the beta period until MLA-C02 reaches general availability. That distinction matters for anyone using an older MLA-C01<\/a> study plan in October 2026.<\/p>\n

The C01 blueprint still matters because it defines the traditional machine-learning-engineering foundation that MLA-C02 expands. MLA-C01 validates the ability to build, operationalize, deploy, and maintain ML solutions and pipelines on AWS. Its four scored domains are Data Preparation for ML at 28%, ML Model Development at 26%, Deployment and Orchestration of ML Workflows at 22%, and ML Solution Monitoring, Maintenance, and Security at 24%.<\/p>\n

English candidates should not treat MLA-C01 as the live exam anymore<\/h3>\n

AWS explicitly states that September 28, 2026 was the final day to take MLA-C01 in English. Since September 29, the updated MLA-C02 beta has been available in English only. MLA-C01 remains available in Japanese, Korean, and Simplified Chinese during the beta period.<\/p>\n

That means the correct current decision depends on language. English-language candidates should shift to MLA-C02 preparation, while candidates taking one of the remaining C01 languages can still use the C01 blueprint until AWS retires it at general availability of the updated exam.<\/p>\n

The MLA-C01 format was 65 questions in 130 minutes<\/h3>\n

AWS lists MLA-C01 as an Associate-level exam with 65 questions and 130 minutes. The standard price is USD 150. AWS documentation states that 50 questions affect the score and 15 are unscored, unidentified items used for statistical evaluation. The passing score is 720 on a 100\u20131,000 scaled score.<\/p>\n

The exam guide includes multiple-choice, multiple-response, ordering, and matching question types, while the public certification overview emphasizes multiple choice and multiple response. Unanswered questions are scored as incorrect and there is no penalty for guessing.<\/p>\n

Domain 1 focused on data preparation for machine learning<\/h3>\n

The 28% data-preparation domain covers ingesting and storing data, transforming data and performing feature engineering, and ensuring data integrity before modeling. AWS names formats such as Parquet, JSON, CSV, ORC, Avro, and RecordIO, along with storage and streaming sources including S3, EFS, FSx, Kinesis, Kafka, and Flink.<\/p>\n

Transformation topics include cleaning, outliers, imputation, scaling, standardization, binning, encoding, tokenization, feature engineering, labeling, and AWS services such as AWS Glue<\/a>, DataBrew, SageMaker Data Wrangler, and Feature Store.<\/p>\n

Bias and data governance appeared before model training<\/h3>\n

MLA-C01 also expects candidates to recognize class imbalance, selection bias, measurement bias, data quality problems, anonymization, masking, encryption, personally identifiable information, protected health information, and data residency concerns.<\/p>\n

This is a useful reminder that ML engineering begins with trustworthy data. A model can be tuned perfectly and still be operationally or ethically weak if the training data is biased, invalid, or handled in violation of policy.<\/p>\n

Domain 2 covered model selection, training, tuning, and evaluation<\/h3>\n

The 26% model-development domain asks candidates to choose general modeling approaches, train and refine models, tune hyperparameters, analyze model performance, and manage versions. Candidates should understand common algorithm families, the relationship between problem type and model choice, and the metrics appropriate for classification, regression, or other tasks.<\/p>\n

Amazon SageMaker<\/a> is central because the exam expects familiarity with building, training, tuning, and managing ML workloads using AWS-native capabilities rather than only knowing generic machine-learning theory.<\/p>\n

Evaluation required matching metrics to business risk<\/h3>\n

Accuracy is not always the right metric. Precision, recall, F1, ROC\/AUC, RMSE, MAE, and confusion-matrix behavior can matter differently depending on whether false positives or false negatives are more expensive.<\/p>\n

The exam also expects candidates to detect overfitting or underfitting and to compare training and validation behavior rather than optimizing one number without understanding generalization.<\/p>\n

Domain 3 connected deployment with infrastructure and CI\/CD<\/h3>\n

The 22% deployment and orchestration domain includes selecting deployment infrastructure, scripting infrastructure, endpoints, compute resources, auto scaling, and automated CI\/CD. It connects model engineering with DevOps and MLOps.<\/p>\n

AWS CloudFormation<\/a>, Step Functions<\/a>, and AWS CodePipeline<\/a> are useful internal references because the domain expects reproducible infrastructure and automated ML workflows rather than one-off manual deployments.<\/p>\n

Deployment choice followed latency, scale, and cost requirements<\/h3>\n

Real-time endpoints, asynchronous inference, batch inference, serverless options, containers, and autoscaling represent different operating models. The right deployment depends on request rate, acceptable latency, payload size, cost, traffic variability, and whether inference can run offline.<\/p>\n

A review of model deployment and optimization on AWS<\/a> can help candidates connect model behavior with runtime infrastructure and scaling choices.<\/p>\n

Domain 4 covered monitoring, maintenance, security, and cost<\/h3>\n

The final 24% domain includes monitoring model inference, detecting drift, monitoring infrastructure and costs, troubleshooting workflows, securing resources, encryption, access control, and compliance. SageMaker Model Monitor, SageMaker Clarify, CloudWatch, IAM, KMS, and related AWS controls are part of the operating model.<\/p>\n

CloudWatch<\/a>, IAM<\/a>, and KMS<\/a> reinforce the distinction among observability, authorization, and encryption. Production ML requires all three.<\/p>\n

The transition to MLA-C02 changes scope, not the importance of C01 foundations<\/h3>\n

MLA-C02 adds generative AI, foundation models, Amazon Bedrock, RAG, agentic AI, and responsible-AI content. AWS also shifts domain weights to 28%, 24%, 24%, and 24%. Traditional C01 skills in data preparation, model training, deployment, CI\/CD, monitoring, and security remain relevant, but English candidates should now study them through the updated C02 blueprint.<\/p>\n

The transition creates an important editorial distinction between certification validity and exam availability. A credential earned through MLA-C01 remains valid for its normal three-year period even though the English exam itself has closed. AWS explicitly tells candidates who already passed C01 that their certification remains active. Retirement of an exam version does not invalidate credentials already earned under that version.<\/p>\n

MLA-C02 broadens the role because machine-learning engineers increasingly work with foundation models, Amazon Bedrock, retrieval-augmented generation, agentic workflows, and responsible-AI concerns in addition to classical ML. AWS keeps the four-domain structure but shifts Domain 2 from 26% to 24% and Domain 3 from 22% to 24%, with Domains 1 and 4 at 28% and 24% respectively.<\/p>\n

C01’s data-preparation scope remains highly relevant to C02 because even foundation-model applications depend on trustworthy data pipelines, storage, labeling, transformation, bias awareness, and security. English candidates should not discard C01 knowledge; they should add the newer GenAI and FM objectives on top of it.<\/p>\n

Storage trade-offs are especially important in Domain 1. S3 is a common object store, but EFS, FSx, EBS, relational databases, DynamoDB, streaming services, and feature stores can all appear depending on training, inference, or pipeline requirements. The exam expects candidates to choose from access pattern, throughput, persistence, cost, and data structure.<\/p>\n

Feature Store is an example of where data engineering and ML engineering meet. Reusable, governed features can reduce duplicate transformation logic between teams and improve consistency between training and inference. The underlying idea is not only \u201cknow the service name,\u201d but understand why feature reuse and online\/offline access matter to production ML.<\/p>\n

Model versioning also connects C01’s development and deployment domains. Training jobs can produce several candidates; evaluation selects an acceptable model; a model registry or version-management workflow preserves artifacts and metadata; deployment automation promotes the chosen version. A production incident should be traceable back to the exact model and pipeline state that produced it.<\/p>\n

Infrastructure selection in Domain 3 includes endpoints, compute, containers, and autoscaling because model serving is a systems problem. A low-latency online API, a large offline batch prediction, and an asynchronous image workload can use different deployment patterns even when they serve the same model family.<\/p>\n

Security in Domain 4 crosses the entire lifecycle. Data at rest and in transit needs protection; training roles should use least privilege; model artifacts and endpoints need controlled access; network placement can limit exposure; logs and audit evidence support compliance. Security should not be memorized as one end-of-exam list.<\/p>\n

Cost optimization is also part of operations. Training instances, endpoint uptime, storage class, batch versus online inference, autoscaling, idle resources, and repeated data processing can materially change spend. The best ML engineering design meets performance and reliability needs without leaving expensive capacity unused.<\/p>\n

For editorial and study purposes in October 2026, the safest wording is therefore precise: MLA-C01 is the previous English blueprint and remains temporarily available in Japanese, Korean, and Simplified Chinese during the C02 beta period. MLA-C02 is the updated English beta route. Any preparation material that simply calls C01 \u201cthe current English exam\u201d is now stale.<\/p>\n

AWS’s target-candidate guidance also helps explain the level. MLA-C01 assumes at least one year of experience with SageMaker and related AWS ML services and familiarity from roles such as backend development, DevOps, data engineering, MLOps, or data science. It is not a data-science-theory exam and it is not an enterprise architecture exam; it validates implementation and operational skills around ML systems.<\/p>\n

The public guide explicitly lists full end-to-end ML-solution architecture, broad ML strategy, deep specialization across multiple ML domains, and advanced quantization analysis among out-of-scope tasks. That boundary is useful when studying: candidates should know enough architecture to deploy and operate within an existing design, but they are not being tested as principal ML architects.<\/p>\n

The C01 service list is non-exhaustive and subject to change. That means study should be organized around tasks\u2014ingest, transform, train, deploy, orchestrate, monitor, secure\u2014rather than memorizing a static list of AWS names. New or renamed services can be placed more easily when the engineering function is understood.<\/p>\n

For candidates continuing C01 in non-English languages, use AWS’s current status page to confirm that the selected language remains bookable at the time of registration. The beta transition is temporary, and C01 availability will end when MLA-C02 reaches general availability across the supported languages.<\/p>\n

Within the broader AWS certification<\/a> path, MLA-C01 is best treated as a legacy\/current-language-specific blueprint during the beta transition. Its content is still valuable, but exam-version and language status must be checked before scheduling.<\/p>\n","protected":false},"excerpt":{"rendered":"

MLA-C01 is now in transition rather than being the current English exam. AWS ended English delivery of MLA-C01 on September 28, 2026, and opened the English-only MLA-C02 beta on September 29. MLA-C01 continues in Japanese, Korean, and Simplified Chinese during the beta period until MLA-C02 reaches general availability. That distinction matters for anyone using an […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[1648,1647],"tags":[],"_links":{"self":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/26327"}],"collection":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/comments?post=26327"}],"version-history":[{"count":1,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/26327\/revisions"}],"predecessor-version":[{"id":26328,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/posts\/26327\/revisions\/26328"}],"wp:attachment":[{"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/media?parent=26327"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/categories?post=26327"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.examlabs.com\/certification\/wp-json\/wp\/v2\/tags?post=26327"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}