AWS Certified AI Practitioner Study Guide

The AWS Certified AI Practitioner credential validates a broad, conceptual understanding of artificial intelligence, machine learning, and generative AI as they apply to the AWS Cloud. Rather than testing hands-on model-building skills, it focuses on whether a candidate can explain core AI/ML terminology, recognize appropriate use cases, and understand how AWS services support AI workloads responsibly and securely.

This exam sits at the entry point of the AWS AI/ML certification track, positioned below specialty credentials like AWS Certified Machine Learning Engineer. It is designed for a wide audience rather than deep technical specialists alone.

Who Should Take It

• Business analysts, product managers, and sales professionals who need to speak credibly about AI capabilities on AWS
• Developers and IT professionals who are new to AI/ML and want a structured on-ramp
• Technical leaders who evaluate AI vendor and architecture decisions but don't build models directly

Why It Matters for Your Career

As organizations race to adopt generative AI, employers increasingly want proof that staff outside the data science team understand foundational concepts like large language models, prompt engineering, and responsible AI practices. Holding this certification signals fluency in AI terminology and AWS's AI service portfolio, which can differentiate a candidate in roles spanning cloud consulting, product management, and technical sales. It also serves as a credible stepping stone toward more advanced AWS machine learning certifications later in a career path.

Understanding the mechanics of the exam helps you plan your study timeline and testing-day logistics with confidence.

Format and Timing

• The exam consists of 65 questions in total
• Of those, 50 questions are scored and contribute to your result
• The remaining 15 questions are unscored and used by AWS to evaluate future exam content, though you won't know which ones they are while testing
• You are given 90 minutes to complete the exam

Scoring

The exam uses a compensatory scoring model, meaning your overall performance is what counts rather than needing to pass each content domain individually — a strong showing in one domain can offset a weaker one elsewhere. The minimum passing score is 700 on a scaled range of 100 to 1,000. Results are typically available within five business days after you finish the exam.

Cost and Delivery

The exam costs 100 USD, consistent with AWS's pricing for Foundational-level exams. You can take it either at a Pearson VUE testing center or as an online proctored exam through OnVUE, which allows you to test from a private space like your home or office. Scheduling happens through Pearson VUE from your AWS Certification Account, and you can cancel or reschedule up to 24 hours before your appointment without incurring additional fees.

Certification Validity

Once earned, the certification is valid for 3 years, after which recertification is required to keep the credential active.

The exam is organized into 5 content domains, each weighted differently in terms of how much of the scored content it represents. Knowing these weightings helps you allocate study time proportionally rather than spreading effort evenly across topics that matter unequally.

Domain 1: Fundamentals of AI and ML (20%)

This domain covers basic AI, ML, and deep learning concepts, including terminology like training, inference, supervised versus unsupervised learning, and common use cases where AI/ML delivers business value versus where traditional programming is more appropriate.

Domain 2: Fundamentals of Generative AI (24%)

The largest foundational domain, this covers core generative AI concepts such as foundation models, large language models, tokens, embeddings, prompt engineering basics, and the business considerations around adopting generative AI, including advantages and limitations.

Domain 3: Applications of Foundation Models (28%)

As the most heavily weighted domain, this focuses on how to design, use, and customize foundation model-based solutions in practice — covering techniques like prompt engineering, retrieval-augmented generation (RAG), fine-tuning considerations, and relevant AWS services such as Amazon Bedrock.

Domain 4: Guidelines for Responsible AI (14%)

This domain addresses responsible AI development, including fairness, bias, explainability, and transparency considerations that shape how AI systems should be designed and evaluated.

Domain 5: Security, Compliance, and Governance for AI Solutions (14%)

The final domain covers securing AI systems, data privacy, regulatory compliance considerations, and governance practices needed to operate AI solutions responsibly within an organization.

Because this is a Foundational-level exam covering conceptual breadth rather than deep technical depth, most candidates with some cloud or tech background can prepare effectively in four to six weeks of steady, part-time study. Here's a topic-by-topic approach that mirrors the domain weightings.

Weeks 1-2: Build the Foundation

Start with Domain 1 and Domain 2 together, since GenAI concepts build on general ML fundamentals. Focus on terminology: supervised versus unsupervised learning, training versus inference, foundation models, tokens, embeddings, and prompt engineering basics. Use flashcards to drill vocabulary early, since much of this exam rewards precise recall of terms.

Weeks 3-4: Go Deep on Applications

Since Domain 3 carries the heaviest weighting, dedicate the most study time here. Learn how Amazon Bedrock and related AWS AI services fit together, how retrieval-augmented generation works conceptually, and when fine-tuning versus prompt engineering versus RAG is the appropriate solution design choice. Work through scenario-based practice questions that mimic the exam's applied style.

Week 5: Responsible AI and Governance

Cover Domains 4 and 5 together, since both deal with the guardrails around AI systems: bias, fairness, explainability, data privacy, and compliance. These domains are lighter in weighting but still commonly tested with scenario questions.

Week 6: Review and Practice Exams

Spend the final stretch taking full-length practice exams under timed conditions, reviewing the glossary for any terms that still feel unfamiliar, and revisiting weaker domains identified through practice results. Prioritize understanding why wrong answers are wrong, not just memorizing correct ones.

Before Exam Day

• Confirm your testing method in advance — whether at a Pearson VUE testing center or via OnVUE online proctoring — since each has different check-in requirements and technical setup steps
• If testing online, test your webcam, microphone, and internet connection well ahead of time to avoid last-minute proctoring issues
• Review your identification requirements, since testing centers and online proctoring both enforce strict ID verification

During the Exam

• Pace yourself against the clock: with 65 questions in the allotted time, budgeting roughly a minute or so per question leaves room to flag and revisit harder items
• Since scoring is compensatory, don't panic over a handful of tough questions in one domain — a strong overall performance is what matters, not perfection in every section
• Watch for scenario-based questions that describe a business problem and ask you to pick the best-fit AWS service or technique; eliminate obviously wrong answers first
• Remember that some questions are unscored and used for future exam calibration, so don't waste time trying to guess which ones don't count — treat every question with equal care

Common Mistakes to Avoid

• Over-studying deep technical implementation details rather than focusing on conceptual understanding, which is what this Foundational-level exam actually tests
• Neglecting the responsible AI and governance domains because they carry lower individual weighting — combined, they still make up a meaningful share of the exam
• Skipping practice questions in favor of passive reading, which often leaves gaps in applying concepts to scenario-style questions
• Forgetting to check cancellation or rescheduling policies if your plans change, since deadlines apply if you need to shift your appointment

Preparing for a broad, terminology-heavy exam like this one benefits from a mix of study formats, and this site's free resources are built to cover each stage of that process.

Practice Questions

Scenario-based practice questions mirror the applied style used throughout the real exam, particularly for the heavily-weighted Applications of Foundation Models domain. Working through these regularly helps you get comfortable identifying the best-fit AWS service or technique in a business scenario, rather than just recalling isolated facts.

Flashcards

Given how much of this exam rests on precise terminology — foundation models, tokens, embeddings, fine-tuning, RAG, and more — flashcards are an efficient way to drill vocabulary in short repeated sessions. They're especially useful during the early weeks of study when you're still building a working vocabulary for generative AI concepts.

Glossary

A comprehensive glossary lets you quickly look up unfamiliar terms as you encounter them in practice questions or outside study materials, reinforcing definitions in context rather than in isolation. This is particularly helpful for candidates newer to AI/ML who need a fast reference without switching to outside sources.

Combining these three formats — active recall through flashcards, applied reasoning through practice questions, and quick lookups through the glossary — supports the kind of well-rounded conceptual understanding this Foundational-level exam is designed to test.