AWS Machine Learning Specialty
Amazon Web Services
Complete guide to passing the AWS Machine Learning Specialty exam on your first attempt.
$300
~50%
3 years
Global
Amazon Web Services
$150k-$200k
Are you ready for AWS Machine Learning Specialty?
Loading quiz...
Complete Overview
AWS Certified Machine Learning - Specialty (exam code MLS-C01) was Amazon's 300 USD specialty certification for people who built, trained, tuned and deployed machine learning models on AWS. AWS retired it: the last day to sit the exam was 31 March 2026, and the exam prep resources on AWS Skill Builder were withdrawn on the same date. Anyone who already holds it keeps an active certification for three years from the date it was earned, and can continue displaying the Credly digital badge for that period.
While it ran, MLS-C01 was a 180-minute exam of 65 questions delivered at a Pearson VUE test centre or through online proctoring, in English, Japanese, Korean and Simplified Chinese. Only 50 of those 65 questions counted. AWS included 15 unscored items, unmarked and mixed into the paper, to trial future questions. Results came back as a scaled score between 100 and 1,000 with a minimum passing score of 750, using a compensatory model in which the overall score decides the outcome and no individual section has to be passed.
The blueprint split four ways and summed to 100 percent: Data Engineering at 20 percent, Exploratory Data Analysis at 24 percent, Modeling at 36 percent and Machine Learning Implementation and Operations at 20 percent. AWS described the target candidate as someone with two or more years of experience developing, architecting and running ML or deep learning workloads in the AWS Cloud, performing an AI and ML development or data science role.
The replacement AWS points people to is AWS Certified Machine Learning Engineer - Associate, exam code MLA-C01. It costs 150 USD, runs 130 minutes, contains 65 questions of which 50 are scored, and has a minimum passing score of 720 on the same 100 to 1,000 scale. Its four domains are Data Preparation for ML at 28 percent, ML Model Development at 26 percent, Deployment and Orchestration of ML Workflows at 22 percent, and ML Solution Monitoring, Maintenance, and Security at 24 percent. It also introduces two question types MLS-C01 never used: ordering items, where three to five responses must be placed in the correct sequence, and matching items, where every pair must be matched correctly against three to seven prompts.
MLS-C01 spent 36 percent of its weight on modeling, including framing business problems, algorithm selection, training, hyperparameter optimisation and evaluation. MLA-C01 pushes weight toward operations: CI/CD pipelines, endpoint auto scaling, drift monitoring, cost management and IAM least privilege. If you were studying for the specialty exam and missed the retirement date, most of your data preparation and model development work carries over, and the gap is in deployment, orchestration, monitoring and security.
Why Get AWS Machine Learning Specialty Certified?
Holders keep an active certification for three years from the date it was earned, and AWS confirms no action is required from them despite the retirement. The Credly digital badge stays shareable for that whole period.
The MLS-C01 blueprint put 36 percent of the exam on Modeling, the heaviest single-domain weight of any AWS certification in the ML family, covering problem framing, algorithm selection, training, hyperparameter optimisation and evaluation.
The exam ran 180 minutes for 65 questions, roughly 2.8 minutes per item, which gave more room for long scenario questions than the 130 minutes MLA-C01 allows for the same question count.
AWS gives every certification holder a 50 percent discount voucher toward their next exam, accessible from the Benefits section of the AWS Certification Account. That halves the 150 USD cost of moving to MLA-C01.
The four MLS-C01 domains map cleanly onto MLA-C01: Data Engineering and Exploratory Data Analysis feed Data Preparation at 28 percent, and Modeling feeds ML Model Development at 26 percent, so prior study is not wasted.
Qualifying US veterans covered under a GI Bill education provision can claim reimbursement from Veterans Affairs for AWS Certification exam fees when the exam is purchased through the AWS Certification Account.
AWS publishes both exam guides in full, including in-scope and out-of-scope service lists, so you can audit which AWS services the successor exam expects you to know and which it rules out, from AWS DeepRacer and Amazon Monitron through to Amazon Cognito and AWS WAF.
Exam Format & Structure
Duration
180 minutes. The successor exam, MLA-C01, runs 130 minutes.
Questions
65 questions. AWS states that 50 of them affect your score and 15 are unscored trial items that are not identified on the exam.
Passing Score
750 out of a scaled range of 100 to 1,000. AWS uses a compensatory scoring model, so you pass on the overall score and do not need to reach a threshold in each section. MLA-C01 uses the same scale with a minimum passing score of 720.
Question Types
- Multiple choice: one correct response and three distractors
- Multiple response: two or more correct responses out of five or more options, with no partial credit
- Note that MLA-C01 adds ordering items of three to five responses and matching items against three to seven prompts, neither of which awards partial credit
Delivery Method
Pearson VUE test centre or online proctored exam. Retired on 31 March 2026, after which neither option can be booked. Results are posted within five business days to the AWS Certification Account under Exam History.
Exam Domains & Topics
The heaviest domain, spanning five objectives from business problem framing through evaluation. AWS asked candidates to justify an ML approach rather than merely name one, which meant recognising when a problem was not an ML problem at all. Hyperparameter optimisation was scoped to basic tuning, and complex algorithm development was listed as out of scope.
Key Topics to Master:
- Framing business problems as machine learning problems
- Selecting appropriate models for supervised, unsupervised and reinforcement problems
- SageMaker built-in algorithms including XGBoost, Linear Learner, BlazingText, Object2Vec, DeepAR and Random Cut Forest
- Training with script mode against TensorFlow, PyTorch and MXNet
- Basic hyperparameter optimisation with SageMaker automatic model tuning
- Regularisation, early stopping and overfitting or underfitting diagnosis
- Evaluation metrics including confusion matrix, precision, recall, F1, ROC and AUC, and RMSE
- Choosing the right metric for imbalanced classification rather than defaulting to accuracy
Three objectives covering sanitisation, feature engineering and visualisation. This domain rewarded knowing which AWS tool to reach for at which data scale, since the same cleaning task has a Glue answer, an EMR answer and a pandas-in-a-notebook answer, and the exam expected the one that fits the stated volume and latency constraints.
Key Topics to Master:
- Handling missing data through imputation, deletion and indicator variables
- Detecting and treating outliers
- Feature engineering: binning, log transformation, scaling, standardisation and normalisation
- Encoding: one-hot, label and binary encoding, plus tokenisation for text
- Dimensionality reduction with PCA and t-SNE
- Data labelling with Amazon SageMaker Ground Truth and Amazon Mechanical Turk
- Visualisation with Amazon QuickSight and notebook plotting libraries
- Identifying class imbalance before it reaches training
Three objectives covering repositories, ingestion and transformation. The exam treated the data lake as the default ML repository and expected fluency in the streaming stack, particularly which Kinesis service applies to which pattern. Questions frequently turned on cost and format choice rather than on whether an approach would work at all.
Key Topics to Master:
- Creating data repositories on Amazon S3, including partitioning and lifecycle policies
- Choosing between Kinesis Data Streams, Data Firehose and Kinesis Video Streams
- Batch ingestion with AWS Glue crawlers and jobs
- Transformation with AWS Glue, Amazon EMR with Spark, and AWS Lambda
- Columnar formats: Parquet and ORC versus row formats such as CSV and JSON
- RecordIO-protobuf as the efficient input format for SageMaker built-in algorithms
- Querying with Amazon Athena over data catalogued in the AWS Glue Data Catalog
- Choosing between S3, EFS and FSx as a training data source
Four objectives covering production characteristics, service selection, security and deployment. The service-selection objective was where the AI services mattered: knowing that a given business need is solved by Comprehend, Rekognition, Transcribe, Translate, Personalize or Forecast rather than by a custom model. Advanced networking and advanced DevOps concepts were listed as out of scope.
Key Topics to Master:
- Designing for performance, availability, scalability, resiliency and fault tolerance
- Multi-AZ endpoint deployment and endpoint auto scaling
- A/B testing with SageMaker production variants
- Choosing between real-time endpoints and batch transform
- Selecting the right managed AI service: Comprehend, Rekognition, Transcribe, Translate, Polly, Textract, Personalize, Forecast, Lex
- Basic AWS security: IAM roles for SageMaker, VPC endpoints, KMS encryption at rest and in transit
- SageMaker Neo for edge deployment and Elastic Inference for accelerator attachment
- Monitoring with Amazon CloudWatch and logging with AWS CloudTrail
Recommended Study Plan
- 1Check whether you already hold MLS-C01 and when it expires, since it is active for three years from the date earned
- 2Download the AWS Certified Machine Learning Engineer - Associate (MLA-C01) exam guide, which is the exam AWS now points ML candidates toward
- 3Read the appendix of in-scope and out-of-scope AWS services and mark every service you cannot describe in one sentence
- 4Claim your 50 percent discount voucher from the Benefits section of your AWS Certification Account if you already hold any AWS certification
- 5Enrol in the MLA-C01 Exam Prep Plan on AWS Skill Builder and take the Official Practice Question Set to get a baseline
- 1Load the same dataset into S3 as CSV, then as Parquet, and compare Athena query cost and runtime between them
- 2Build a Kinesis Data Streams to Data Firehose to S3 pipeline and confirm records land in the expected partitions
- 3Ingest data into SageMaker Data Wrangler and push engineered features into SageMaker Feature Store
- 4Merge two sources with AWS Glue and again with Spark on EMR, and note when each is the cheaper answer
- 5Compare S3, EBS, EFS and FSx as training data sources against the tradeoffs listed in the exam guide
- 1Run a cleaning job in AWS Glue DataBrew and reproduce the same transformations in Data Wrangler
- 2Apply scaling, binning, log transformation and one-hot encoding, then measure the effect on a baseline model
- 3Use AWS Glue Data Quality to validate a dataset and fail a rule deliberately
- 4Run SageMaker Clarify pre-training bias metrics and interpret class imbalance and difference in proportions of labels
- 5Label a small dataset with SageMaker Ground Truth and inspect the output manifest
- 1Train the same problem three ways: a SageMaker built-in algorithm, script mode with PyTorch, and a JumpStart foundation model
- 2Run SageMaker automatic model tuning with random search, then with Bayesian optimisation, and compare the trials
- 3Deliberately overfit a model, then apply dropout and weight decay and record the change in validation metrics
- 4Register three model versions in SageMaker Model Registry and practise promoting one
- 5Use SageMaker Clarify to explain predictions and SageMaker Debugger to diagnose a convergence problem
- 1Build a confusion matrix on an imbalanced dataset and show why accuracy misleads where F1 does not
- 2Plot an ROC curve and read AUC, then compare against a precision-recall curve on the same data
- 3Compare a shadow variant against a production variant on a SageMaker endpoint
- 4Practise selecting between a custom model and an AI service such as Comprehend, Rekognition or Textract for stated business needs
- 5Work through the cost, latency and accuracy tradeoff questions in the AWS Skill Builder practice question set
- 1Deploy one model four ways: real-time endpoint, serverless endpoint, asynchronous endpoint and batch transform
- 2Configure endpoint auto scaling on invocations per instance and prove it scales under load
- 3Build a SageMaker Pipelines workflow, then rebuild the same flow with AWS Step Functions
- 4Set up a CodePipeline, CodeBuild and CodeDeploy chain that retrains and redeploys on a Git push
- 5Practise blue/green, canary and linear rollout strategies and know which rollback each supports
- 1Configure SageMaker Model Monitor and trigger a data quality violation on purpose
- 2Use SageMaker Clarify to detect a distribution shift and distinguish data drift from concept drift
- 3Right-size an endpoint with SageMaker Inference Recommender and AWS Compute Optimizer
- 4Apply a resource tagging strategy and track endpoint spend in AWS Cost Explorer with a budget alert
- 5Write a least-privilege IAM policy for a SageMaker execution role and deploy an endpoint inside a VPC
- 1Take the AWS Certification Official Pretest and score it by domain against the 28, 26, 22 and 24 percent weights
- 2Practise ordering and matching items specifically, since these formats did not exist on MLS-C01 and give no partial credit
- 3Read the MLA-C01 out-of-scope service appendix end to end once, so you can eliminate a distractor on sight
- 4Take the AWS Certification Official Practice Exam and require a clear margin over 720 before booking
- 5Book MLA-C01 at 150 USD through your AWS Certification Account, applying the 50 percent discount voucher if you hold one
Ready to pass AWS Machine Learning Specialty?
Get 500+ practice questions, video walkthroughs, and a pass guarantee.
Best Study Resources
AWS Certified Machine Learning - Specialty (MLS-C01) exam guide
Official exam guideThe retired exam's official guide, still published in AWS documentation. Version 2.4 contains the four domain weights, the 750 passing score, the 50 scored and 15 unscored item split, and the service lists that named AWS Data Pipeline, AWS DeepRacer and Amazon Machine Learning as out of scope.
Free
AWS Certified Machine Learning Engineer - Associate (MLA-C01) exam guide
Official exam guideThe successor exam's guide, with four domains at 28, 26, 22 and 24 percent, the 720 passing score, the four question types, the 50 scored plus 15 unscored split and a full appendix of in-scope and out-of-scope AWS services.
Free
AWS Skill Builder Exam Prep Plan for MLA-C01
Structured digital courseAWS's own four-step preparation plan: exam-style questions, knowledge refresh, review and practice, then a readiness assessment. Includes the Official Practice Question Set, the Official Pretest and the Official Practice Exam.
Free tier plus paid subscription for full features
AWS Builder Labs
Hands-on lab environmentGuided labs in a live AWS account, referenced directly in the AWS exam prep steps. The way to practise endpoint deployment, auto scaling and pipeline orchestration without running up a personal bill.
Paid subscription
Amazon SageMaker Developer Guide
Product documentationThe primary source for built-in algorithm input formats, hyperparameters, endpoint types, Model Monitor, Clarify, Feature Store and Pipelines. Most exam distractors are resolved by one paragraph of this guide.
Free
AWS Certification FAQs
Policy referenceThe authority on the 14-day retake wait, the two-year block on retaking a passed exam, five-business-day results, the 24-hour cancellation window, the two-reschedule limit and the 50 percent discount voucher.
Free
AWS Certification Information and Policies
Policy referenceCovers identification requirements, requesting accommodations before registering, beta exam rules and the teen testing policy for candidates aged 13 to 17.
Free
AWS Machine Learning Blog
Technical blogWhere AWS publishes the reference architectures that scenario questions are drawn from, including drift detection patterns, multi-model endpoint designs and cost optimisation for inference.
Free
AWS Well-Architected Machine Learning Lens
WhitepaperAWS's own design guidance for ML workloads. The MLA-C01 exam guide names ML lens design principles explicitly under the monitoring task statement, which makes this required rather than optional reading.
Free
Common Mistakes to Avoid
Still preparing for MLS-C01. The last day to sit it was 31 March 2026, and AWS withdrew the exam prep resources on the same date.
Switch to MLA-C01. AWS names it as the offering for people who wanted the specialty certification, and it costs 150 USD against the specialty exam's 300 USD.
Assuming a MLS-C01 certification earned shortly before retirement is worth less. It is not, and AWS states no action is needed from holders.
The credential stays active for three years from the date it was earned and the Credly badge stays shareable for that window. When it approaches expiry, MLA-C01 is the exam to sit instead, since MLS-C01 cannot be retaken.
Carrying MLS-C01 study habits into MLA-C01 without adjusting for question types. The specialty exam used only multiple choice and multiple response.
Drill ordering and matching items specifically. Ordering requires three to five responses in the correct sequence, matching requires every pair correct against three to seven prompts, and neither awards partial credit.
Budgeting time as though MLA-C01 gives the same room as MLS-C01. Both have 65 questions, but the specialty exam allowed 180 minutes and the associate exam allows 130.
That is 2 minutes per question instead of 2.8. Practise under a 130-minute clock, and note that ordering and matching items take longer to work than a four-option multiple choice, so elimination speed matters more than it did on the specialty exam.
Treating the 15 unscored questions as a reason to skip hard items. They are not identified on the exam, so you cannot know which ones they are.
Answer everything. AWS scores unanswered questions as incorrect and applies no penalty for guessing, so a considered guess strictly beats a blank on both MLS-C01 and MLA-C01.
Studying algorithm mathematics in depth. Both exam guides list complex mathematical proofs and extensive algorithm development as out of scope.
Learn which algorithm fits which problem shape, what input format it needs and which metric evaluates it. For MLA-C01, spend the recovered hours on Domain 3 and Domain 4, which together carry 46 percent.
Ignoring the appendix of out-of-scope services in the MLA-C01 guide. It names dozens of services, including AWS DeepRacer, Amazon Monitron, AWS Panorama, Amazon Cognito and AWS WAF.
Read both lists once. Recognising that a service is out of scope lets you eliminate a distractor immediately in a question where two answers look equally plausible.
Waiting until the last minute to request testing accommodations. AWS arranges these with Pearson VUE before registration, not after.
Request accommodations through the process on the AWS Certification policies page before you book. Note also that AWS offers an ESL accommodation adding 30 minutes when the exam is taken in English.
Rescheduling repeatedly. AWS states that its certification exams can be rescheduled only twice.
Book a date you can keep. Cancellation or rescheduling is free up to 24 hours before the appointment; inside that window you lose the fee entirely, and a no-show forfeits it with no refund.
Exam Day Tips
- 1
Confirm which exam you are actually booked for. MLS-C01 can no longer be scheduled after 31 March 2026, so any current booking is for MLA-C01 at 130 minutes rather than 180.
- 2
Answer every question. AWS scores unanswered items as incorrect and applies no penalty for guessing, and 15 of the 65 items do not count toward your score anyway.
- 3
Multiple response items award no partial credit. If a question asks for two correct answers out of five, three right and one wrong scores the same as nothing.
- 4
Watch for the short service names. AWS states that MLA-C01 uses abbreviated AWS service names and provides a mapping to full names through the Help button inside the exam.
- 5
Do not expect a result on screen. Most AWS exams do not display pass or fail at the end, and detailed results appear in your AWS Certification Account within five business days.
- 6
Your Credly badge email may arrive before your score does. AWS notes this ordering explicitly, so a badge notification is the earliest signal that you passed.
- 7
For an online proctored sitting, the exam is delivered by Pearson VUE, so the OnVUE system check, room scan and clear-desk rules apply rather than any AWS-specific process.
- 8
If English is not your first language and you are testing in English, request the ESL accommodation in advance for an extra 30 minutes. AWS also offers the written exam in eight languages with an English, Japanese or Spanish speaking proctor.
- 9
If you fail, note the section-level table on your score report. AWS uses a compensatory model, so the fix is usually one weak domain rather than an even shortfall across all four.
Career Paths & Salary Ranges
Machine learning engineer
Builds and operationalises ML pipelines in production. AWS positions MLA-C01 for exactly this role, describing it as validating technical ability in implementing ML workloads in production and operationalising them.
$150k-$200k
MLOps engineer
Owns the CI/CD, retraining and monitoring layer around models. MLA-C01 gives Deployment and Orchestration 22 percent and Monitoring, Maintenance and Security 24 percent, so nearly half the exam is this job.
$150k-$200k
Data scientist on AWS
AWS lists data scientist among the candidate roles for MLA-C01 and described MLS-C01 as suited to an AI and ML development or data science role. The modeling and evaluation content maps to daily work.
$150k-$200k
Data engineer moving into ML
AWS names data engineer as a candidate role for MLA-C01. Data Preparation for ML is the largest domain at 28 percent, which makes it the shortest path across for someone already fluent in Glue, Kinesis, S3 and Athena.
$150k-$200k
ML solutions architect
Designs end-to-end ML systems against cost, latency and compliance constraints. Note that AWS listed designing and architecting full end-to-end ML solutions as out of scope for MLA-C01, so this role usually pairs the certification with the Solutions Architect track.
$150k-$200k
Prerequisites & Requirements
- None were required. AWS stated that no specific certification had to be earned first, though candidates commonly held AWS Certified Solutions Architect - Associate, AWS Certified Machine Learning Engineer - Associate or AWS Certified Data Engineer - Associate before attempting the specialty exam.
- AWS described the ideal candidate as having two or more years of experience developing, architecting and running ML or deep learning workloads in the AWS Cloud.
- MLS-C01 can no longer be booked. The last delivery date was 31 March 2026.
- For the successor exam MLA-C01, AWS recommends at least one year of experience with Amazon SageMaker and other AWS ML engineering services, plus one year in a related role such as backend developer, DevOps developer, data engineer or data scientist.
- Candidates must be 13 years or older. Those aged 13 to 17 need consent from a parent or legal guardian, and AWS publishes a Teen Testing Policy covering the full requirements.
Frequently Asked Questions
Is AWS Certified Machine Learning - Specialty still available?
No. AWS retired it and the last day to take the exam was 31 March 2026. The exam preparation resources on AWS Skill Builder were withdrawn on the same date. AWS points candidates to AWS Certified Machine Learning Engineer - Associate (MLA-C01) as the current ML certification, and to AWS Skill Builder digital training on machine learning.
I already hold MLS-C01. Does it still count?
Yes. AWS states that no action is required and that the certification remains active for three years from the date it was earned. You can continue displaying and sharing the Credly digital badge throughout that period. When it expires you cannot recertify by retaking MLS-C01, so the practical route is MLA-C01.
What did the exam cost, and does the fee vary by region?
300 USD, the standard AWS price for professional-level and specialty exams. Associate exams are 150 USD and AWS Certified Cloud Practitioner is 100 USD. AWS publishes local currency pricing in Australian dollars, euros, Korean won, Japanese yen and Chinese yuan, updated at least annually to reflect exchange rates with a minimum of 30 days notice. Taxes such as VAT may apply on top.
What was the passing score and how does the score scale work?
750 on a scaled range of 100 to 1,000. Scaled scoring equates results across exam forms that differ slightly in difficulty, so the raw number of correct answers needed can vary between deliveries. AWS uses a compensatory model, meaning you pass on the overall score and do not need to reach a threshold in any individual domain. The successor exam MLA-C01 uses the same scale with a passing score of 720.
What happened if you failed?
You waited 14 calendar days before you were eligible to retake it, with no cap on the number of attempts, and you paid the full 300 USD registration fee each time. Your score report contained a table of section-level classifications showing relative strength across the four domains, which AWS advises interpreting with caution because section-level samples are small.
How long do results take?
Within five business days of completing the exam, posted to your AWS Certification Account under Exam History. AWS emails you when they are available. Most AWS exams do not display a pass or fail result on screen at the end, and the Credly digital badge notification sometimes arrives before the results themselves are posted.
What identification do I need for an AWS exam?
AWS exams are delivered by Pearson VUE, so Pearson VUE identification rules apply at both test centres and online proctored sittings: a valid government-issued ID with a recognisable photo, with a name that matches the exam booking exactly. AWS publishes its identification requirements on the AWS Certification Information and Policies page.
What are the online proctoring rules?
Online proctoring is available for all AWS certification exams when you schedule with Pearson VUE. The session uses Pearson VUE's OnVUE service, which requires a system check in advance, a room scan at check-in, a clear desk and a private space with no other people present. Reference materials, notes and calculators are not permitted.
Can I use a calculator or reference materials?
No. No reference materials, notes or personal calculators are permitted at a test centre or during an online proctored sitting. The one in-exam aid AWS documents for MLA-C01 is a mapping of abbreviated AWS service names to their full names, reachable through the Help button inside the exam.
Are testing accommodations available?
Yes. AWS states that it provides reasonable accommodations for individuals with disabilities and that these are arranged with the test delivery provider before you register for the exam. Pearson VUE does not share accommodation request details with AWS, so documentation goes to the provider. AWS separately offers an ESL accommodation adding 30 minutes when you take the written exam in English.
How does recertification work for AWS certifications?
AWS certifications are valid for three years and you recertify by passing the latest version of the same exam before it expires. Because MLS-C01 no longer exists, its holders cannot follow that path. AWS also gives every certification holder a 50 percent discount voucher toward their next exam, accessible from the Benefits section of the AWS Certification Account, which applies to recertifying or to a different exam.
Can I cancel or reschedule an AWS exam?
Yes, up to 24 hours before your scheduled appointment with no additional fee, managed through the Manage Pearson VUE exams button in your AWS Certification Account. Inside that 24-hour window you cannot cancel or reschedule. Missing the appointment forfeits the fee with no refund, does not count as a fail, and blocks you from registering again for 24 hours. AWS allows an exam to be rescheduled twice.
How does MLS-C01 compare with MLA-C01?
MLS-C01 was a specialty exam of 180 minutes at 300 USD, weighted toward modeling at 36 percent, aimed at people with two or more years of ML on AWS. MLA-C01 is an associate exam of 130 minutes at 150 USD, weighted toward operations, with Deployment and Orchestration at 22 percent and Monitoring, Maintenance and Security at 24 percent, aimed at people with roughly one year of SageMaker experience. MLA-C01 also adds ordering, matching and case study question types.
How does it compare with AWS Certified AI Practitioner?
AI Practitioner is a foundational credential aimed at people who need to understand AI and ML concepts and AWS AI services without building production pipelines. MLA-C01 is the technical implementation exam. If your job is to ship models and keep them running, MLA-C01 matches; if your job is to make decisions about AI adoption, the foundational route fits better.
What AWS services did the specialty exam expect me to know?
Amazon SageMaker across its full surface, plus the data stack around it: S3, Kinesis Data Streams and Data Firehose, AWS Glue, Amazon EMR, Athena, Redshift and QuickSight, and the managed AI services including Comprehend, Rekognition, Transcribe, Translate, Polly, Textract, Personalize, Forecast and Lex. Advanced networking, advanced database work, advanced DevOps concepts and EMR DevOps tasks were listed as out of scope.
Does passing an AWS exam block me from retaking it?
Yes, for two years. AWS states that once you have passed an exam you cannot retake the same exam for two years, unless the exam has been updated with a new exam guide and a new exam series code, in which case you are eligible for the new version.
Can US veterans get the exam fee reimbursed?
Yes. Any qualifying US veteran covered under a GI Bill with an education provision can submit a reimbursement request to Veterans Affairs for AWS Certification exams taken after 10 December 2015. You must purchase the exam through your AWS Certification Account and submit the VA application. Only AWS Certification exam fees qualify, not training costs.
How current are the AWS services on an exam?
A new product, service or feature must be generally available for three months before it appears on any certification exam. AWS also publishes exam guide revisions at least one month before the changes reach the exam itself. That lag is why an exam guide dated some months ago is still the correct study target.
Pass AWS Machine Learning Specialty, Guaranteed
94% pass rate on first attempt
One-time • Lifetime access