MLA-C02 prerequisites: what you need first

There are no formal prerequisites for MLA-C02 — no required certification, nothing checked at booking. But AWS publishes a target candidate description that is unusually specific, and it is the best predictor of whether you will pass.

What AWS expects

• At least 1 year of experience using Amazon SageMaker AI, Amazon Bedrock and other AWS services for ML engineering
• At least 1 year in a related role: backend software developer, DevOps developer, data engineer or data scientist
• Experience with both traditional ML and generative AI

Plus general IT knowledge:

• Common ML algorithms and their use cases
• Foundation model capabilities, limitations and use cases
• Data engineering fundamentals — formats, ingestion, transformation
• Querying and transforming data
• Software engineering practices: modular code, deployment, debugging
• Provisioning and monitoring cloud resources
• CI/CD pipelines and infrastructure as code

And AWS knowledge:

• SageMaker AI capabilities and algorithms, for traditional ML and generative models
• Amazon Bedrock features
• AWS storage and data processing services
• Deploying applications and infrastructure on AWS
• Monitoring and logging tools
• AWS services for CI/CD automation
• Security practices: IAM, encryption, data protection

Which of these actually matter

Not all gaps are equal. In rough order of how much they will hurt:

1. No hands-on SageMaker AI or Bedrock. This is the critical one. The exam is written by people assuming you have deployed an endpoint and called a model. Close this before anything else.
2. No CI/CD or infrastructure as code experience. A quarter of the exam is deployment and orchestration, and pipelines are central to it.
3. No production ML operations experience. Drift, monitoring and cost optimisation are hard to learn from reading because they are about what goes wrong over time.
4. Weak on data engineering. The largest domain at 28%, but more learnable from study than the others.
5. No generative AI exposure. C02 covers foundation models throughout. Narrower than it looks, and quick to close with Bedrock.

What you do not need

• No certification. Not AIF-C01, not a cloud practitioner credential.
• No data science depth. You are not being examined on the mathematics of algorithms.
• No architecture seniority. AWS explicitly puts designing full end-to-end solutions and setting ML strategy out of scope.
• No deep specialisation. Working deeply in two or more ML domains is also out of scope.

If you are short of the year

Be honest about it. This exam is difficult to pass on reading alone, and the questions are scenario-based in a way that rewards having seen production behave badly.

Two reasonable paths:

Build something real. One project taking a dataset through preparation, training, deployment to an endpoint, a CI/CD pipeline and monitoring covers all four domains. A few months of that beats a year of tutorials.

Take AIF-C01 first. Foundational, 100 USD, no prerequisites. It will not teach you engineering, but it builds the vocabulary and is a fair ramp if your experience is closer to six months than twelve.

Cost of preparation

You need an AWS account you can build in. Keep an eye on it: endpoints bill while provisioned, training jobs bill while running, and GPU instances are expensive. Delete endpoints when you finish a session — the habit costs nothing and is also examinable content.

Before you book

1. Deploy a model to an endpoint and call it.
2. Run one training job, and tune one hyperparameter.
3. Put a pipeline around a deployment.
4. Set up monitoring and look at what it reports.
5. Try the free sample questions.

If step 1 is unfamiliar territory, you are not ready yet — and the fix is a weekend of building, not more reading.