Prepstellar
Get the app

AIF-C01 · Getting started

22 cards

AIF-C01 Exam Overview

Swipe, scroll or use ← →

  1. What the certification validates

    AWS Certified AI Practitioner is a foundational certification for knowledge of AI, machine learning, generative AI concepts, use cases, and AWS AI services. The AIF-C01 exam emphasizes practical business applications: describing AI approaches, selecting appropriate technologies for use cases, and using them responsibly.

    What the exam asks for What that looks like
    Describing AI, ML, and generative AI concepts and methods Naming what a technique is and what it is for
    Identifying appropriate uses of those technologies Matching a business problem to a suitable approach
    Determining the right type of technology for a use case Choosing between options that are all technically real
    Using the technologies responsibly Recognizing fairness, transparency, security, and governance duties
    1 / 22

  2. What the certification validates

    That list explains why this course spends its time connecting definitions to decisions. The exam validates identifying suitable AI/ML and generative AI technologies for business problems, not writing model code.

    A question therefore rarely stops at what is this. It usually sets a scene — a support team buried in tickets, a document archive nobody can search — and asks which approach fits. The definition is the entry ticket; the decision is what earns the point.

    2 / 22

  3. Quick check

    Which task is closest to the depth AIF-C01 is designed to measure?

    1. AWriting the training loop that fine-tunes a model

      Writing model code is explicitly outside what this certification measures.

    2. BChoosing an AI approach for a business problem

      Right. The exam validates identifying suitable AI/ML and generative AI technologies for business problems.

    3. CDeriving the statistics behind a model's error metric

      Mathematical and statistical analysis of models belongs to more specialized certifications.

    3 / 22

  4. Who the exam is written for

    The target candidate has up to six months of exposure to AI/ML technologies on AWS and uses, but does not necessarily build, AI/ML solutions. That is a deliberately wide door: people in cloud, development, data, IT, and line-of-business roles all fit through it.

    Read the second half of that sentence carefully, because it is the part that surprises people. Uses, but does not necessarily build means someone who applies existing services and has never written a training job is exactly who the exam was designed for — not someone who is under-qualified for it.

    4 / 22

  5. Quick check

    A colleague has four months of experience using AI services on AWS, has never built a model, and asks whether the exam is meant for her. What does the candidate description say?

    1. AShe must build and deploy an AI/ML pipeline in production first

      Building pipelines is listed among the tasks the target candidate is not expected to perform.

    2. BShe needs several years of AI/ML infrastructure engineering experience before she can qualify

      No engineering seniority is required; the described exposure is up to six months.

    3. CShe fits it: the profile allows up to six months and covers people who use, not only build

      Right. The target candidate has up to six months of exposure and uses, but does not necessarily build, AI/ML solutions on AWS.

    5 / 22

  6. Who the exam is written for

    The exam guide is equally explicit about what it does not test. Developing algorithms, implementing data or feature engineering, tuning models, and building AI/ML infrastructure are outside the target candidate's expected job tasks.

    Outside the candidate's job tasks Still expected of the candidate
    Developing or coding AI/ML models and algorithms Describing what a model or an algorithm is for
    Data engineering and feature engineering Recognizing why data quality changes results
    Hyperparameter tuning and model optimization Knowing that tuning exists and what it adjusts
    Building AI/ML pipelines and infrastructure Choosing the service that provides them

    The exam still recommends familiarity with core AWS services, the shared responsibility model, IAM, and AWS pricing models. Foundational does not mean contextless.

    6 / 22

  7. Quick check

    Which task does the exam guide place outside the target candidate's expected job tasks?

    1. ADeveloping or coding AI/ML models and algorithms

      Right. Developing or coding models and algorithms is listed among the out-of-scope job tasks.

    2. BIdentifying an AI technology that suits a business use case

      Matching a technology to a use case is one of the abilities the exam validates.

    3. CRecognizing the responsible use of AI/ML technologies

      Responsible use is a scored domain of its own, so it is firmly inside the scope.

    7 / 22

  8. Keep your progress in the app

    That’s 3 of 10 quick checks. In the app they stay answered, and every lesson remembers where you left off.

    Get the app Keep reading here

  9. Five domains and their weights

    The exam organizes scored content into five domains, and the weights are published in advance.

    Domain Share of scored content
    Fundamentals of AI and ML 20%
    Fundamentals of GenAI 24%
    Applications of Foundation Models 28%
    Guidelines for Responsible AI 14%
    Security, Compliance, and Governance for AI Solutions 14%

    Fundamentals of AI and ML accounts for 20 percent, Fundamentals of GenAI for 24 percent, Applications of Foundation Models for 28 percent, Guidelines for Responsible AI for 14 percent, and Security, Compliance, and Governance for AI Solutions for 14 percent.

    8 / 22

  10. Five domains and their weights

    Applications of Foundation Models is the largest domain, but the exam uses all five areas to assess foundational judgment across concepts, business use, responsibility, and security.

    Two readings of that table are worth keeping apart:

    • Where the questions are. Generative AI and foundation models together carry more than half the scored content, so hours spent there pay back fastest.
    • Where the risk is. The two 14 percent domains are small but never optional. Responsible AI and security also surface inside scenarios that belong to the other three domains.
    9 / 22

  11. Quick check

    You have a fixed number of study hours. What do the published weights actually tell you?

    1. AOnly Applications of Foundation Models is scored, so the other four domains can safely wait

      Every one of the five domains contributes scored questions, so none of them can be skipped.

    2. BThe five domains are equally weighted, so the hours split evenly

      The weights differ: they run from 14 percent up to 28 percent.

    3. CApplications of Foundation Models leads at 28 percent, and all five domains are scored

      Right. Applications of Foundation Models is the largest domain, and the exam still assesses all five areas.

    10 / 22

  12. Questions, formats, and time

    The exam presents 65 questions in 90 minutes. That averages a little under 85 seconds per question: comfortable for a definition, tight for a long scenario.

    The practical consequence is a pacing rule. Read the scenario, answer it, and move on. Minutes spent re-reading an early question are minutes missing from a later one, and the later one is worth exactly the same.

    11 / 22

  13. Quick check

    How much time does the exam leave per question on average?

    1. AAbout two minutes, from 50 questions in 100 minutes

      The exam is not 50 questions long; 50 is the number of questions that affect the score.

    2. BA little under 85 seconds, from 65 questions in 90 minutes

      Right. AIF-C01 presents 65 questions in 90 minutes, which averages just under 85 seconds each.

    3. CAbout three minutes, from 30 questions in 90 minutes

      No published figure puts the exam at 30 questions, and the pacing would be far looser than the real one.

    12 / 22

  14. Questions, formats, and time

    Four item formats can appear, and they are not read the same way.

    Format How it works
    Multiple choice One correct response and three incorrect responses
    Multiple response Two or more correct responses among five or more options; all of them are needed for credit
    Ordering Three to five responses placed in the correct sequence
    Matching Responses paired with three to seven prompts; every pair must be right

    Multiple choice asks which single option is best. The other three ask for a complete answer, and an answer that is almost complete earns nothing at all.

    13 / 22

  15. Quick check

    A question offers six options and states that more than one is correct. What earns credit?

    1. ASelecting all of the correct responses, since this format gives no partial credit

      Right. Multiple response requires every correct response out of five or more options before the item counts.

    2. BSelecting any one correct response, since a single right pick is enough

      Partial selections earn nothing on this format, however reasonable the chosen option is.

    3. CPlacing all six options into the correct sequence, since six options signals an ordering item

      Ordering items present three to five responses to sequence, which is a different format.

    14 / 22

  16. Scored, unscored, and unanswered

    Fifty questions affect the score, while 15 unscored questions are not identified during the exam. They look exactly like every other item, so no question can safely be treated as one worth skipping.

    Unanswered questions are incorrect, and there is no penalty for guessing. Those two rules together settle every hesitation on the day.

    Situation What the rules imply
    You know the answer Answer it and move on
    You are down to two plausible options Pick one; a blank scores the same as a wrong answer
    You have no idea Answer anyway, because guessing costs nothing
    15 / 22

  17. Quick check

    Two minutes are left, one question is still blank, and the answer is a coin flip. What do the scoring rules imply?

    1. ALeave it blank, because an unanswered item is treated more gently

      A blank is scored as incorrect, so leaving it gains nothing over a guess.

    2. BAnswer it, because unanswered items are incorrect and guessing is not penalized

      Right. Unanswered questions are scored as incorrect and there is no penalty for guessing.

    3. CSkip it, because questions reached late in the exam are the unscored ones

      The 15 unscored questions are not identified during the exam and are not tied to any position.

    16 / 22

  18. How the result is reported

    Results use a scaled score from 100 to 1,000, with 700 as the minimum passing score. Scaled scoring lets results be compared across exam forms that may differ slightly in difficulty, so 700 means the same standard whichever form a candidate sits.

    The scoring model is compensatory: a candidate passes the exam overall and does not need a passing score in every section. A weaker domain can be carried by stronger ones. That is why the weights guide how study time is spent, while the result itself depends on overall performance. The score report may also show section-level classifications, and those are feedback rather than separate pass marks.

    17 / 22

  19. Quick check

    A candidate performs well in four domains, poorly in one, and finishes above 700 overall. What does the compensatory model mean here?

    1. AThe weakest domain decides the outcome on its own

      No single section can fail a candidate whose overall performance meets the standard.

    2. BEach individual domain has to clear 700 before any overall score is calculated at all

      Section-level results are reported as feedback, not as separate thresholds to clear.

    3. CThe overall score passes, because no single section needs its own passing mark

      Right. The compensatory model means only the overall exam has to be passed.

    18 / 22

  20. Prepare at the intended depth

    The official preparation path starts with learning the exam, refreshing AWS knowledge and skills, reviewing and practicing each domain, and assessing readiness. Each step answers a different question: what is on it, what am I missing, does the practice hold up, and am I ready.

    If AWS Cloud itself is new, the certification page recommends beginning with AWS Cloud Practitioner Essentials or AWS Technical Essentials. This course supplies an AI/ML orientation, but the exam's recommended AWS familiarity remains part of readiness — the core services, the shared responsibility model, IAM, and the pricing models still show up inside AI scenarios.

    19 / 22

  21. Quick check

    Someone entirely new to AWS asks what to do before working through the AI content. What does the certification page recommend?

    1. AStart with AWS Cloud Practitioner Essentials or Technical Essentials

      Right. Candidates new to AWS Cloud are pointed to Cloud Practitioner Essentials or Technical Essentials first.

    2. BSkip cloud fundamentals, since this exam only covers AI and ML topics

      Familiarity with core services, the shared responsibility model, IAM, and pricing is part of the recommended knowledge.

    3. CStart by tuning model hyperparameters, since practice builds AWS familiarity

      Hyperparameter tuning is one of the tasks placed outside the target candidate's expected job tasks.

    20 / 22

  22. Key takeaways

    • AIF-C01 is foundational and targets people who use, but do not necessarily build, AI/ML solutions on AWS.
    • The exam has 65 questions in 90 minutes; 50 are scored and 15 unidentified questions are unscored.
    • Question formats can include multiple choice, multiple response, ordering, and matching.
    • The passing score is 700 on a 100-to-1,000 scale, and only the overall exam must be passed.
    • Applications of Foundation Models has the largest domain weight at 28 percent.

    Everything above points the same way: study for judgment across all five domains, answer every question, and aim at the overall standard rather than a perfect section.

    21 / 22

  23. Quick check

    Which summary of the exam facts is accurate?

    1. A65 questions in 65 minutes, every one of them scored, passing at 650

      The exam runs 90 minutes, 15 of its questions are unscored, and the passing score is 700.

    2. B65 questions in 90 minutes, 50 of them scored, passing at 700 on a 100-to-1,000 scale

      Right. Those are the published duration, question count, scored count, and passing score.

    3. C50 questions in 90 minutes, 15 marked as practice, passing at 1,000

      The exam presents 65 questions, the unscored ones are not identified, and 1,000 is the top of the scale rather than the pass mark.

    22 / 22

  24. 10 quick checks · then the test

    In the app, finishing the quick checks opens this lesson’s 10-question test, and the ones you miss come back exactly when you’re about to forget them.

    Take this lesson’s test → Next lesson: AI and ML Core Terminology and ParadigmsGet the app
← Previous Keep your progress · Get the app Next card →

The whole course, on your phone

Lessons you can read, audio you can listen to on the way to work, and practice that remembers what you got wrong.

Get the app