Lesson 1: How AIP-C01 Works and How to Use This Course

Why start with the exam itself

Before you write any Bedrock code, you need to know what you're being tested on. AWS publishes an exam guide for AIP-C01. It's the official syllabus. It lists the question types, the scoring rules, the five content domains with their weightings, and every task and skill that's in scope. This course follows that guide. When anything in a level differs from the guide, the guide is right.

Some ground rules for the course:

• The guide belongs to AWS and isn't reproduced here. It's free on the AWS Certification site, at the AWS Certified Generative AI Developer - Professional exam page and in AWS documentation as "Exam Guide (AIP-C01)". Download it now and keep it open while you study.
• This course is independent of AWS and isn't endorsed by AWS. No course can promise you a pass, and this one doesn't.
• Check the AWS exam page for duration, price and languages. As of October 2026, AWS lists 180 minutes, 300 USD, and English, Japanese, Korean and Simplified Chinese, with testing at a Pearson VUE test centre or online with a proctor. These details change, so check them on the day you book.

Who AWS has in mind

The guide describes the target candidate as someone with:

• 2 or more years of experience building production-grade applications on AWS or with open-source technologies
• general AI/ML or data engineering experience
• 1 year of hands-on experience implementing GenAI solutions

It also lists five areas of recommended AWS knowledge:

1. Experience with AWS compute, storage and networking services
2. Understanding of AWS security best practices and identity management
3. Experience with AWS deployment and infrastructure as code (IaC) tools
4. Familiarity with AWS monitoring and observability services
5. Understanding of AWS cost optimization principles

Here's a quick self-assessment. Run it and be honest with your scores.

# readiness.py - rate yourself 0 (never used) to 3 (use it in production)
recommended = {
    "Compute, storage, networking (Lambda, EC2, S3, VPC)": 2,
    "Security best practices and IAM": 2,
    "Deployment and IaC (CloudFormation, CDK)": 1,
    "Monitoring and observability (CloudWatch, X-Ray)": 1,
    "Cost optimization principles": 0,
}
for area, score in recommended.items():
    flag = "  <- gap: review before Domain 3/4 levels" if score < 2 else ""
    print(f"{score}/3  {area}{flag}")

Sample output:

2/3  Compute, storage, networking (Lambda, EC2, S3, VPC)
2/3  Security best practices and IAM
1/3  Deployment and IaC (CloudFormation, CDK)  <- gap: review before Domain 3/4 levels
1/3  Monitoring and observability (CloudWatch, X-Ray)  <- gap: review before Domain 3/4 levels
0/3  Cost optimization principles  <- gap: review before Domain 3/4 levels

Practitioner note: Professional-level questions assume IAM and networking knowledge without explaining it. A question might really be about Bedrock Guardrails, but the right answer can depend on knowing which IAM policy or VPC endpoint is missing. Close those gaps early.

What's out of scope

The guide says these tasks are out of scope for the target candidate. The list isn't exhaustive.

• Model development and training
• Advanced ML techniques
• Data engineering and feature engineering

This is a developer exam, not a data-scientist exam. You'll be tested on integrating foundation models (FMs), RAG, agents, prompts, safety, cost and evaluation. You won't be tested on deriving loss functions. If you notice yourself deep in training hyperparameters, you've probably drifted off the syllabus.

Question types and scoring

The guide describes two question formats:

• Multiple choice: one correct response and three incorrect responses (distractors).
• Multiple response: two or more correct responses out of five or more options. You must select all the correct responses to get credit. There's no partial credit.

The key rules:

• 65 scored questions plus 10 unscored questions, so 75 in total. The unscored ones are being trialled for future exams. They aren't identified, so treat every question as if it counts.
• An unanswered question is scored as incorrect, and there's no penalty for guessing. Never leave a blank.
• Results are pass or fail, reported as a scaled score of 100–1,000. The minimum passing score is 750. Scaling evens out small differences in difficulty between exam forms, so 750 doesn't mean 75% correct.
• Scoring is compensatory. You don't need to pass each section, only the overall exam. A strong Domain 1 can make up for a weak Domain 5. The section-level table on your score report shows strengths and weaknesses, but AWS itself says to interpret it with caution.

Domain weightings, and how this course follows them

This script turns the weightings into approximate scored questions and course levels:

weights = {"D1": 0.31, "D2": 0.26, "D3": 0.20, "D4": 0.12, "D5": 0.11}
SCORED, LEVELS = 65, 59   # 59 content levels after this orientation level
for d, w in weights.items():
    print(f"{d}: ~{w*SCORED:4.1f} scored questions, ~{round(w*LEVELS)} levels")

D1: ~20.1 scored questions, ~18 levels
D2: ~16.9 scored questions, ~15 levels
D3: ~13.0 scored questions, ~12 levels
D4: ~ 7.8 scored questions, ~7 levels
D5: ~ 7.2 scored questions, ~6 levels

These numbers are estimates, not AWS figures. They show that Domains 1 and 2 together make up more than half the exam (57%), and the course spends its time to match.

Gotcha: Don't skip Domains 4 and 5 because they're small. About 15 scored questions come from them, and they're often the easiest marks to gain with targeted study.

How to use this course

1. Keep the official guide open and tick off each skill as the matching level covers it.
2. Run the boto3 examples in a sandbox account with Amazon Bedrock model access enabled.
3. Read the explanation for every quiz distractor. The real exam relies on plausible wrong answers.
4. Book the exam once your weakest domain is reasonably solid. With compensatory scoring, you don't need it to be perfect.

What the exam asks about this

The exam doesn't test its own rules, but knowing them changes how you take it. Answer every question. On multiple-response questions, select exactly as many options as the question asks for. Spend your study time according to the weightings.

Key takeaways

• Get the free AIP-C01 exam guide from AWS and check the AWS exam page for current length, price and languages. This course is independent of AWS and can't promise a pass.
• There are 75 questions: 65 scored and 10 unscored, and you can't tell which is which. Formats are multiple choice (1 of 4) or multiple response (2 or more of 5 or more, all needed for credit).
• Blanks are scored as wrong and there's no penalty for guessing, so answer everything.
• You need 750 on a 100–1,000 scaled score. Scoring is compensatory: pass overall, not per section.
• The domain weights are 31/26/20/12/11%.
• Target candidate: 2 or more years building production apps, AI/ML or data experience, and 1 year of hands-on GenAI. Five recommended AWS knowledge areas.
• Out of scope: model development and training, advanced ML techniques, and data and feature engineering.