THE DEV BENCH
🟧

AWS Machine Learning Engineer – Associate · MLA-C01

One workspace for passing the AWS MLA-C01 — a scenario-based multiple-choice exam about operating ML on AWS (SageMaker at the center): preparing data, building & deploying models, and monitoring + securing them. Four domain decks and an 89-question practice exam do the heavy lifting, authored against the official AWS exam guide.

The honest read: this is an associate exam that tests operating ML on AWS, not the math of ML — it assumes you already know ML fundamentals and basic AWS. Almost every question is “which AWS service fits this scenario,” so the decks + exam here are the core study. Real hands-on SageMaker needs a paid AWS account, so those labs point out honestly to the AWS Free Tier + Skill Builder rather than pretending to host them.

Deep work · AWS MLA-C01total
or log

The exam at a glance

verified July 2026
Cost$150
Format65 questions · multiple choice / multiple response
Duration130 minutes
Passing score720 / 1000
Validity3 years
DeliveryPearson VUE · online or test center

AWS scores associate exams on a scaled 100–1000 range with 720 to pass; the raw question count that maps to is not published, and some items are unscored. Confirm current cost, format, and policy on the official certification page before you book.

Certification roadmap

sequence
CLF-C02AWS Cloud PractitionerOptional foundation

Foundational · MCQ

Not required. Worth it only if AWS itself (IAM, S3, regions, the console) is unfamiliar. Skip if you already work in AWS day to day.

MLA-C01ML Engineer – AssociateThe goal

Associate · 65 Q · 130 min · 720/1000 · 3-year validity · $150

The target of this workspace. SageMaker-centric: data prep, model development, deployment/orchestration, and monitoring + security. It tests OPERATING ML on AWS — which service fits which job — not the math of ML. It assumes you already know ML fundamentals and basic AWS.

TrioTri-cloud MLOpsThe bigger story

AWS MLA · Azure AI-300 · Google PMLE

MLA-C01 is the AWS leg of an 'MLOps across all three clouds' story alongside Azure AI-300 and Google PMLE. Same concepts, three vendors' tooling — a strong portfolio signal. Both other paths have workspaces here too.

Exam domains — weight your study by these

official weights
DomainWeightWhat it covers
Data Preparation for ML28%Ingesting & storing data (S3, Kinesis, Glue), transforming & feature-engineering (Glue, Data Wrangler, DataBrew, EMR, Feature Store), labeling (Ground Truth), and ensuring data integrity, quality & bias (Clarify) — plus encoding, scaling, and split choices.
ML Model Development26%Choosing an approach (built-in algos, JumpStart, Autopilot, or your own container), training & tuning (training jobs, Automatic Model Tuning, distributed & spot training, instance types), and evaluating against the business goal.
Monitoring, Maintenance & Security24%Monitoring models & data drift (Model Monitor, Clarify), observability & audit (CloudWatch, CloudTrail), cost/performance optimization, and securing ML (IAM least-privilege, VPC isolation, KMS, Secrets Manager, governance).
Deployment & Orchestration22%The right inference option (real-time, serverless, async, batch), sizing & auto-scaling, safe rollout (blue/green, canary, A/B), and ML CI/CD with SageMaker Pipelines, Model Registry, Projects, and IaC.

Well balanced — no domain dominates. Data Prep (28%) + Monitoring/Security (24%) together outweigh the modeling itself, which is the point: this is an ENGINEERING cert, not a data-science one. The 98 deck cards are distributed to match these weights.

Flashcard decks — the recall layer

4 decks · 98 cards
Data Preparation for ML (28%)30 cards — the biggest domain. Glue vs EMR, Data Wrangler vs DataBrew, Feature Store (online vs offline), Kinesis ingestion, Athena, Ground Truth, Clarify for bias, and the encoding/scaling/splitting/leakage choices that decide whether a model can learn. Start here; it's over a quarter of the exam.

The other three domains

Practice exam

live · 89-question pool
AWS MLA-C01 — ML Engineer Associate Practice ExamA fresh 50-question draw each attempt from a domain-weighted 89-question pool, scored at 72%to mirror the real 720/1000 cut, with a 130-minute soft timer. Scenario-style “which AWS service” MCQs with worked explanations and a per-domain weak-area breakdown. Progress saves in your browser — no login.

Use the per-domain breakdown to find weak areas, then close them with the matching deck. These are original questions in the exam's scenario style — a strong readiness gauge, not a brain dump of real items.

Hands-on SageMaker — the honest note

external · needs an AWS account

The MLA-C01 is multiple-choice, so you can pass on the decks + exam + reading. But building the muscle memory — actually running a data-prep → train → deploy → monitor loop in SageMaker — needs a real (paid) AWS account, which we can't host for you. Rather than fake it, this path points you at the real thing:

Cost watch:SageMaker endpoints and notebook instances bill by the hour even when idle — the number-one surprise bill. Stop/delete them the moment you're done, and set a billing alarm.

Skill map — where to practice each thing

the workspace
SkillPractice withStatus
Which AWS service fits each job (all domains)The 4 MLA decks (98 cards)ready
Exam-format scenario recall & weak-area findingMLA-C01 practice exam (draw 50, pass 72%)ready
Hands-on SageMaker (data prep → train → deploy)AWS Free Tier + Skill Builder labs (external)external
The official question style & scopeAWS Skill Builder Exam Prep Plan (external, free)external

A study routine

  1. 1.Learn the SageMaker MAP first: for each service, know whether it does data prep, training, deployment, or monitoring. Most questions are 'which AWS service fits this scenario' — the map is the whole game. Do the Data Preparation deck first (28%).
  2. 2.Weight your time by the blueprint: Data Prep (28%) + Monitoring/Security (24%) together outweigh the modeling itself. Those are exactly the parts a pure data-science course skips — don't under-study them.
  3. 3.Drill the 'which inference option' decision until it's reflex: real-time (steady low-latency), serverless (spiky, cold-start OK), async (huge payloads / long jobs), batch (offline bulk). It shows up constantly.
  4. 4.Practice reading scenario stems for the CONSTRAINT — cost, latency, drift, security, or data volume. That single constraint usually picks the answer among several services that would 'work.'
  5. 5.Use the practice exam to find weak domains via the per-domain breakdown, then close them with the matching deck. Hit 72% (the real 720/1000 cut) three times before you book.

Curated resources — free-first

verified July 2026

Everything you need to LEARN the MLA-C01 is free (the official exam guide + Skill Builder prep plan + SageMaker docs). Only the exam voucher ($150) and optional deep labs cost money. Every link checked live (July 2026); prices/versions shift, so treat costs as “as seen.”

Start here (free)

Hands-on & deeper (free tier / optional)

Start the practice exam