Amazon Web Services · 2026 Edition
A complete preparation guide written by Amazon Web Services-certified engineers. Covers the exam format,all 4 blueprint domains, a week-by-week study plan, and proven tips for passing first time.
2–4 months
Prep time
Intermediate–Advanced
Difficulty
50
Exam questions
700/1000
Pass mark
Exam code
MLA-C01
Full name
AWS Certified Machine Learning Engineer Associate
Vendor
Amazon Web Services
Duration
130 minutes
Questions
50 items
Passing score
700/1000 (scaled)
Domains covered
4 blueprint domains
Recommended experience
1+ year of ML or data engineering experience; hands-on SageMaker experience strongly recommended
Typical prep time
2–4 months
The AWS Certified Machine Learning Engineer Associate (MLA-C01) validates practical skills in building, training, deploying, and monitoring ML models on AWS using SageMaker and surrounding services. It bridges the gap between data science and production ML — ideal for ML engineers, data scientists moving into MLOps, and cloud architects building AI platforms.
Job roles this opens
Official Amazon Web Services blueprint weights — study time should roughly match these percentages.
Month 1
Data Engineering for ML: S3 data lakes, Glue ETL, feature engineering, SageMaker Feature Store, data labelling with Ground Truth
Tip: Feature Store is heavily tested — know the difference between online store (low-latency serving) and offline store (training). Know when to use SageMaker Data Wrangler for visual ETL vs Glue for large-scale processing.
Month 2
Model Training & Tuning: SageMaker training jobs, built-in algorithms, hyperparameter tuning, distributed training, Autopilot
Tip: Know SageMaker's built-in algorithms by category: XGBoost/Linear Learner (supervised tabular), K-Means/PCA (unsupervised), BlazingText (NLP/embeddings), DeepAR (time series), Image Classification/Object Detection (CV). The exam asks which algorithm to use for a given problem.
Month 3
Model Deployment & Inference: real-time endpoints, batch transform, serverless inference, multi-model endpoints, A/B testing
Tip: Deployment options matter: real-time endpoint = low-latency API (SageMaker Endpoint); batch transform = offline large-dataset scoring; async inference = large payloads or long processing; serverless = spiky traffic with cold start tolerance. Know when to use each.
Month 4
MLOps & Monitoring: SageMaker Pipelines, Model Registry, Model Monitor, CloudWatch, data drift, concept drift
Tip: SageMaker Model Monitor is a major exam topic. Know: data quality monitoring (detects missing/invalid features), model quality monitoring (detects accuracy drift against ground truth), bias drift (uses Clarify), and feature attribution drift. Know how to set up a monitoring schedule and what baselines are.
SageMaker is the core of this exam. Know the full lifecycle: Data Wrangler → Feature Store → Training Jobs → Hyperparameter Tuning → Model Registry → Endpoints → Model Monitor. Know what each component does and when to use it.
Containers are fundamental: SageMaker uses Docker containers for training and inference. Know when to use AWS pre-built containers (built-in algorithms, framework containers like TF/PyTorch) vs Script Mode (your code in a pre-built container) vs BYOC (fully custom container).
Distributed training: know data parallelism (same model replicated across GPUs, data split between them — use when model fits on one GPU) vs model parallelism (model split across GPUs — use for very large models that don't fit). SageMaker supports both via SageMaker distributed library.
Cost optimisation is tested: SageMaker Savings Plans, Spot Instances for training (use checkpointing to handle interruptions), multi-model endpoints to share infrastructure across models, Inferentia chips for inference cost reduction.
Security: know SageMaker VPC configuration, network isolation, IAM roles for training jobs vs endpoints, S3 encryption at rest (SSE-S3, SSE-KMS), and KMS for SageMaker notebook and endpoint volume encryption.
Apply everything in this guide with adaptive practice questions, detailed answer explanations, and domain analytics.