AIF-C01 study plan: pass in 4 weeks

Four weeks at 5–7 hours a week. This plan assumes some familiarity with AWS but no AI background, and it follows the weighting rather than the domain order — generative AI and foundation models are 52% of the scored content.

You need an AWS account you can experiment in. Bedrock and SageMaker AI appear throughout and are far easier to learn by opening than by reading about.

Week 1: fundamentals of AI and ML (20%)

The vocabulary everything else rests on.

• AI, machine learning and deep learning, and how they relate.
• Supervised, unsupervised and reinforcement learning, and which problems each suits.
• Classification versus regression. Training versus inference.
• Common terms: model, features, labels, overfitting, bias and variance, inference.
• Which AWS service fits which classic problem — recognising the purpose of SageMaker AI and the pre-trained AI services.

Week 2: fundamentals of GenAI (24%)

• Foundation models, large language models, tokens, context windows, embeddings.
• Prompt engineering: zero-shot, few-shot, system instructions, and why the same prompt can give different answers.
• What generative AI is good at and where it fails — hallucination above all.
• Amazon Bedrock: what it is, what it gives you access to, and how it differs from running your own model.
• Where generative AI fits against traditional ML for a given business problem.

Week 3: applications of foundation models (28%)

The largest domain. Spend the most time here.

• Retrieval-augmented generation (RAG) — what it solves, when to use it instead of fine-tuning.
• Fine-tuning and continued pre-training: what each costs and when each is justified.
• Choosing a model: capability against cost against latency.
• Evaluating a foundation model’s output — what “good” means when there is no single right answer.
• Vector stores and embeddings at a conceptual level.
• Agents and multi-step workflows.

Week 4: responsible AI (14%), security and governance (14%), then practice

The two smaller domains together, then revision.

Responsible AI:

• Bias and fairness, transparency and explainability, robustness, and the human role.
• What AWS provides to support these, and why guardrails exist.

Security, compliance and governance:

• The shared responsibility model applied to AI services.
• IAM for controlling access to AI resources.
• Data protection: encryption, and where your prompts and data go.
• Compliance and governance concepts — at the awareness level AWS expects, not implementation.

Then:

• A full practice exam under real conditions: 90 minutes, 65 questions.
• Review every wrong answer, and every right answer you guessed.
• Practise ordering and matching questions specifically — they are all-or-nothing.

Where the hours go

Two things that earn easy marks

Learn the service names properly. A large share of AIF-C01 is recognising which AWS service addresses a described need. Amazon Bedrock, SageMaker AI, and the pre-trained services for vision, speech, text and document processing. You do not need to configure them; you need to know what each is for.

Learn the shared responsibility model applied to AI. It appears in the security domain and it is a concept you can learn in an hour that reliably yields marks.

What not to study

AWS publishes what is out of scope, and it is worth respecting. Do not learn to code a model, tune hyperparameters, build a pipeline or write a governance framework. If your study material is teaching you those, it is preparing you for MLA-C02, not this exam.

Because scoring is compensatory, you pass on the total rather than per domain. If one of the 14% domains stays weak, strength on the 52% generative AI block genuinely compensates.