A certification mentioned on this page has been retired. Microsoft Certified: Azure AI Engineer Associate (AI-102) is no longer available to take. Microsoft reports the retirement date as 2026-06-30. The replacement is Microsoft Certified: Azure AI Apps and Agents Developer Associate (AI-103). Read any reference below as historical, not as advice to take this retired exam. Check the successor's current requirements before planning your preparation.
Some links on this page are affiliate links. If you sign up after clicking one we may earn a commission, at no extra cost to you — and it never affects how we rank or rate anything. How this site is funded.
Quick answer
MLA-C01 — the AWS Certified Machine Learning Engineer Associate — tests the engineering of machine learning on AWS: preparing data, training and tuning models, deploying them, and keeping them monitored, secured and affordable in production. SageMaker sits at the centre of all of it. For an engineer with working Python and some AWS exposure, two to three months of part-time preparation is realistic. Theory helps, but this exam rewards operational fluency — the candidate who has run a pipeline end-to-end beats the one who memorised the algorithms. Exam fee: whatever the provider currently lists.
Where we would actually start
The exam assumes machine-learning fundamentals it does not teach. Sixteen hours of scikit-learn and PyTorch closes that gap faster than re-reading the blueprint.
The hands-on companion to this guide, from Stephane Maarek and Frank Kane — 303 lectures mapped to the same MLA-C01 blueprint.
The table below compares 4 certifications on provider, level, realistic time, coding needed and best for.
| Certification | Provider | Level | Realistic time | Coding needed | Best for |
|---|---|---|---|---|---|
| AWS ML Engineer Associate (MLA-C01) | AWS | Intermediate (associate) | ~2–3 months part-time | Yes (Python) | The exam this guide prepares you for |
| AWS Certified AI Practitioner (AIF-C01) | AWS | Foundational | ~4–6 weeks of prep | No | The lighter on-ramp to AWS AI |
| Machine Learning Specialization | DeepLearning.AI & Stanford Online (Coursera) | Intermediate | ~2–3 months part-time | Yes (Python) | The theory base if your ML fundamentals are missing |
| Google Cloud Professional ML Engineer | Google Cloud | Advanced | ~2–3 months of prep | Yes (Python) | The GCP-side counterpart |
How hard is the MLA-C01 exam?
A genuine step up from the AWS AI Practitioner — this is an associate-level engineering exam, not a literacy check. Scenario questions assume you know how an ML workload actually moves through AWS: where the data lands, what trains the model, how it reaches an endpoint, and what watches it afterwards. Engineers who work near this lifecycle find it demanding but fair; candidates coming from pure theory find the operational framing unfamiliar.
A note on the exam family: AWS's certification lineup around machine learning has been reorganised in recent years, and the long-running ML Specialty exam's status has changed. Check AWS's official certification page for the current lineup before planning a path beyond the associate level.
What's actually on the exam?
Work from the official exam guide — it is the only document that defines scope. The published domains cover, in broad strokes:
- Data preparation for ML — ingesting, transforming and validating data, and choosing the right AWS storage and processing services for it.
- Model development — training, tuning and evaluating models, largely through SageMaker's tooling.
- Deployment and orchestration — getting models into production: endpoints, pipelines, infrastructure choices and automation.
- Monitoring, maintenance and security — drift detection, retraining triggers, cost management, and locking down access to data and models.
Question formats and counts are specified in the exam guide; expect scenario-heavy items where several answers are plausible and one is operationally correct.
Not sure this is the right one for you?
Answer a few questions about your background and what you want the certificate to do, and the picker narrows it to one recommendation — from the same vetted list this page ranks from.
Try the AI Certification Picker →What should you have before starting?
Three things, honestly assessed. Working Python — you will not write code in the exam, but the scenarios assume you could. Basic ML concepts — training/validation splits, overfitting, common model families; if these are new, the Machine Learning Specialization is the right detour before exam prep. And AWS familiarity — you should already know your way around IAM, S3 and the console at roughly the level the AI Practitioner certifies. Missing all three? This is the wrong starting exam; missing one is workable with an extra fortnight.
The eight-week study plan
Built for roughly an hour a day alongside a job, with hands-on work from the first week:
- Weeks 1–2: data preparation. Work through the official exam guide's first domain with the console open — land data in S3, transform it, and meet the ingestion services the scenarios lean on.
- Weeks 3–4: model development. Train and tune something real in SageMaker, twice — once clicking through, once scripted. The mistakes you make here are the exam's distractor answers.
- Weeks 5–6: deployment and monitoring. Stand up an endpoint, build a small pipeline, break it, and watch what the monitoring surfaces. Keep notes; they become revision material.
- Week 7: security and cost. IAM boundaries around ML resources, encryption at rest and in transit, and the cost levers — the least-loved domain and the easiest marks to lose.
- Week 8: the official practice material, full-length and timed. Book the real exam only when you clear it comfortably; otherwise spend a focused week on your weakest domain first.
Why SageMaker hours are the real syllabus
Because the exam's difficulty lives in operational judgment, and judgment only comes from contact. A scenario asking which deployment option fits a latency-sensitive, spiky-traffic workload is trivial if you have deployed both options and painful if you have only read about them. Budget real console time: AWS's free-tier and promotional credits cover meaningful practice, and one end-to-end project — ingest, train, deploy, monitor — touches every domain at once.
The same project does double duty. Written up honestly in a repository, it is interview evidence no multiple-choice score can match — the exam gets you past the screen, the project carries the conversation.
Registration, cost and recertification
Register through AWS's certification portal for online proctoring or a test centre; the fee is set by AWS, and AWS certifications carry an expiry with a recertification cycle. Factor one practice-exam pass and a possible retake into the budget — retake policies and waiting periods are published on the same portal.
Who should take MLA-C01 — and who shouldn't?
Take it if you are an engineer whose work touches the ML lifecycle on AWS — data engineers moving toward model operations, software engineers building ML-backed services, and MLOps-curious platform engineers. Skip it if you cannot code — this exam has no no-code path — or if your employer runs a different cloud: the why AI-102 was retired covers the Azure counterpart and the GCP ML Engineer guide covers Google's, and certifying the cloud you will never touch helps nobody.
Where most MLA-C01 advice gets it wrong
It preps the wrong exam. Advice recycled from theory-heavy ML certifications sends candidates off to grind algorithm derivations — and MLA-C01 barely cares. This is an engineering exam: the questions reward knowing what breaks in production and which AWS lever fixes it, not reciting how gradient descent converges. The candidates who fail with strong ML theory almost always failed the operational half.
The second error is the familiar one from every exam in this cluster: brain dumps. Beyond violating the certification agreement, dumps are self-defeating for an exam whose entire market value is that holders can actually run ML on AWS. If you can pass only with memorised questions, the first on-call incident will disclose that — publicly, and with your name on the deployment.
Verdict
MLA-C01 is the right exam for engineers who want a credential that matches how ML actually ships on AWS — and the wrong exam for anyone hoping to avoid the console. Give it two to three months, spend most of that time hands-on in SageMaker, and let the prep project double as portfolio. If you are earlier in the journey, start with the AI Practitioner study guide and work up. For where this exam sits in the wider stack, see our 2026 rankings and the staged path in the AI certification roadmap — or let the Picker place you.
Ready to start?
Machine Learning Fundamentals in Python — DataCamp · Intermediate · ~16 hrs · subscription. The same option this page recommends above, so you do not have to scroll back for it.
Check price & enrol on DataCamp →Certifications featured in this guide
Every option below is one we cover in depth. Links go to the course on Coursera; where we’ve published a full review, read it first.
Ready to start?
Included in a DataCamp subscription rather than bought outright, so the cost is what you pay while you are working through it — which is an argument for finishing.
Frequently asked questions
How long does it take to prepare for MLA-C01?
Two to three months part-time for an engineer who already writes Python and has some AWS exposure, and longer if either the ML fundamentals or the AWS basics are missing. The variable that moves the timeline most is hands-on hours: candidates who train and deploy something in SageMaker every week prepare faster than those who only watch content, because the exam tests operational judgement rather than recall.
The eight-week plan on this page assumes roughly an hour a day with the console open from week one — data preparation first, then model development, then deployment and monitoring, then the least-loved security and cost domain, then timed practice. If you are missing one of the three prerequisites, add a fortnight. If you are missing all three, this is the wrong exam to start with: the AI Practitioner or a theory course comes first.
Do I need the AI Practitioner before the ML Associate?
No. AWS sets no formal prerequisite, and an engineer already comfortable in the console can prepare for MLA-C01 directly without losing anything. The two exams are aimed at different people rather than stacked in a required sequence — AIF-C01 certifies literacy for non-engineers, MLA-C01 certifies that you can engineer an ML workload on the platform.
AIF-C01 is a useful on-ramp in one specific case: AWS itself is new to you. It teaches the service vocabulary and the shape of the platform cheaply and without code, which makes the associate scenarios much easier to read. If you already know your way around IAM, S3 and the console, skip it — spend the same weeks on SageMaker hours instead, which is where the associate exam's passing margin is actually built.
Does MLA-C01 require coding?
You will not write code in the exam itself — it is question-based, with scenario stems and multiple-choice answers. But it assumes a practitioner-level familiarity that in practice only comes from having written Python against AWS services: the scenarios describe pipelines, training jobs and endpoints as someone who has built them would describe them, and the distractor answers are the mistakes people make while building them.
So there is no realistic no-code path to passing, and none to using the credential afterwards either. If you want a proctored AWS AI credential without code, the AI Practitioner is the exam designed for that and it is a legitimate destination in its own right. If you want this one, the honest prerequisite list is working Python, basic ML concepts, and enough AWS to navigate IAM and S3 without a tutorial.
What is the passing score for MLA-C01?
AWS publishes the scoring model and the passing score in the official exam guide, and that is the only place worth reading it — the score is scaled rather than a raw percentage, and AWS revises it. Read it with the domain weightings in the same document, because the weightings are what should shape your revision.
In practice the more useful gauge is the official practice material. Clearing it comfortably, across every domain rather than on average, matters more than the numeric threshold: one weak domain can sink an otherwise passing performance, and a strong overall score hides exactly that. Book the real exam when your domain-level scores are consistently above the line, and treat a single good run as a coincidence rather than as readiness — by then you may simply be recognising the questions.
Does the AWS ML Engineer Associate certification expire?
Yes. AWS certifications run on a recertification cycle, so plan on renewing rather than passing once and forgetting — and factor that recurring cost, in both money and study time, into what the credential is worth to you. Retake policies, waiting periods and the current cycle are published on AWS's certification portal; check them there rather than relying on a figure from a forum.
There is a genuine upside to it at this level. An associate ML exam is testing a platform that changes fast, and recertification forces your knowledge to track it — the alternative is a credential that says you understood SageMaker as it existed three years ago. Renewal is only a burden if you have stopped doing the work; if you are still building on AWS, it is mostly a scheduled refresh of things you already know.
Keeping this current. Course formats, prices, and certification exam fees change and vary by region. We review our guides regularly — this one was last updated in July 2026 — and we always recommend confirming the specifics on the provider's official page before you enrol.