AWS Certified AI Practitioner (AIF-C01)

About the AIF-C01 exam

AWS Certified AI Practitioner is a foundational exam about using artificial intelligence well, not building it. It is meant for people who work with AI and generative AI on AWS, or decide where to use them, without necessarily training models themselves: product owners, analysts, business and technical leads, and developers new to the field. The target candidate has up to six months of exposure to AI and ML on AWS.

The exam checks that you understand the vocabulary and can match a technique or service to a business problem. Typical questions ask whether a task needs classification, regression or clustering, whether to use prompt engineering, retrieval-augmented generation or fine-tuning, what a setting like temperature does, or how to keep a chatbot from leaking personal data. The exam guide is explicit that coding models, tuning hyperparameters and building ML pipelines are out of scope.

Exam format

• Questions: 65, of which 50 are scored and 15 are unscored trial questions you can't identify. Besides multiple choice and multiple response, the exam can include ordering and matching questions, and each one is all-or-nothing.
• Time: 90 minutes.
• Passing score: 700 on a scaled score from 100 to 1,000.
• Cost: 100 USD, at a Pearson VUE test center or online with a remote proctor.
• Validity: 3 years.
• Languages: AWS has announced that the Italian and German versions retire after October 15, 2026. The exam continues in its other languages.

What the exam covers

The official exam guide has five domains:

• Fundamentals of AI and ML (20%). The difference between AI, ML and deep learning; supervised, unsupervised and reinforcement learning; the ML lifecycle; and metrics such as accuracy, precision, recall and F1, along with overfitting and underfitting.
• Fundamentals of GenAI (24%). Tokens, embeddings, transformers and foundation models; what generative AI is good and bad at, including hallucination; and the cost and capability trade-offs of the services that run these models, chiefly Amazon Bedrock and SageMaker AI.
• Applications of Foundation Models (28%). Choosing a model, inference parameters, prompt engineering techniques, retrieval-augmented generation with knowledge bases and vector stores, fine-tuning, agents, and how to evaluate a model's output.
• Guidelines for Responsible AI (14%). Bias, fairness, explainability, transparency and robustness; human review of low-confidence results; and tools like Bedrock Guardrails and model documentation.
• Security, Compliance, and Governance for AI Solutions (14%). IAM and least privilege for model access, encryption and private connectivity, logging model invocations, prompt injection, data governance and compliance evidence.

How the practice questions here are organised

There are 10 practice tests of 10 questions each, and every test draws on all five domains, so each one samples the whole exam. The real exam's ordering and matching questions can't be shown on this site, so the skills they test are asked as multiple choice instead, for example picking the correct order of steps from four options. Every question explains the right answer and why each other option is wrong, with a link to the AWS documentation behind it.

The full practice exam follows the real one: 65 questions in 90 minutes, weighted by domain the same way, with answers hidden until you submit. It scores all 65 and passes you at 70%, since AWS doesn't publish how the scaled score maps to correct answers.

A study plan that works

• Take Test 1 cold to find your weak domains before reading anything.
• Get the core terms straight first. Many questions turn on one word, such as the difference between fine-tuning and RAG, or between precision and recall.
• For every scenario, ask what the cheapest change is that solves the problem. A better prompt usually beats RAG, and RAG usually beats fine-tuning, unless the scenario rules it out.
• Spend time on the responsible AI and security domains. Together they are 28% of the exam and are easy to underprepare.
• Finish the tests, revisit any you scored under 70% a few days later, then sit the full practice exam timed.

These questions are original, written from the public AWS documentation and the official exam guide. They are not real exam questions, and this site is not affiliated with AWS.