AWS Certified auto_stories Free Compressed Course : 20% preview

AWS Certified AI Practitioner AIF-C01 Certification Course

bolt Everything you need to pass : in one free course.

73 expert modules derived from 65+ exam-style questions. Covers every domain and scenario : organized by blueprint weight so you study what matters most.

Full access from $ 29 One-time · No subscription

play_arrow Start Learning Free payments See Plans

check_circle 15 of 73 modules free · No account needed

Modules

65+

Questions

star star star star star

4.9/5

description Also available: 3-Page Cheat Sheet by Experts

200+ AWS Certified 93% First-Attempt Pass Rate 4.9/5 Rating

About This Course

AWS Certified AI Practitioner AIF-C01 · 73 modules

This course covers every domain tested on the AWS Certified AI Practitioner AIF-C01 exam. Based on our 65+ real practice questions and prepared by certification experts.

info What you'll learn:

Every exam domain with detailed explanations
Common exam traps that catch unprepared candidates
Key concepts, syntax, and configurations
Real-world scenarios aligned with exam objectives
Quick-reference cheat sheets for last-minute review

Your AWS Certified AI Practitioner AIF-C01 Roadmap

AWS Certified AI Practitioner AIF-C01 certification preparation infographic

You're viewing 15 of 73 free modules

The remaining 58 modules cover advanced topics, exam traps, and scenarios that appear on the certification exam.

Unlock All : $ 29

Exam Overview
Exam Domains
Start-to-Finish Study Path
Core Concepts by Domain
Service Selection Guide
Architecture Patterns
Exam Traps and Elimination Rules
Quick Memory Rules
Final Revision Notes
Exam-Day Checklist
Official References

1. Exam Overview

1.1 What the certification validates

AWS Certified AI Practitioner (AIF-C01) is a foundational certification. It tests whether you can:

Explain AI, machine learning (ML), generative AI (GenAI), foundation model (FM), and agentic AI concepts.
Match a business problem to the most suitable AI technique or AWS service.
Recognize when a traditional deterministic solution is better than AI.
Compare traditional ML with foundation-model approaches.
Select sensible GenAI patterns such as prompt engineering, Retrieval Augmented Generation (RAG), fine-tuning, model distillation, and agents.
Recognize responsible-AI, security, compliance, and governance controls.
Reason about tradeoffs: quality, latency, cost, explainability, safety, operational overhead, and regulatory requirements.

This exam is not a coding exam. You are not expected to implement ML algorithms, tune advanced models, or build infrastructure from scratch. Most questions are business scenarios that ask for the best-fitting service, approach, metric, or control.

1.2 Exam format

The current exam guide states:

Item	Details
Scored questions	50
Unscored questions	15
Total questions presented	65
Passing score	700 on a 100–1,000 scale
Scoring model	Compensatory: you pass based on the overall score
Question styles	Multiple choice, multiple response, ordering, matching
Guessing	No penalty for guessing

Practical consequence: answer every question. You do not need to be perfect in every domain, but Domains 2 and 3 deserve extra attention because they represent more than half of the scored exam.

1.3 How to think during the exam

For each scenario, identify five things:

Business outcome: classification, forecast, summary, recommendation, search, automation, document extraction, security control, audit evidence, or governance.
Data type: tabular, text, image, audio, documents, embeddings, private enterprise knowledge, or user interactions.
Response-time need: nightly, interactive, long-running asynchronous, or unpredictable intermittent traffic.
Risk level: public content, regulated decision, sensitive data, user-facing answer, or agent with tool access.
Simplest suitable AWS option: prefer the managed service or lightweight pattern that directly meets the stated requirement.

A common mistake is choosing the most powerful or fashionable option. The exam rewards the smallest correct solution, not the most complex architecture.

2. Exam Domains

Domain	Official weight	Main focus
Domain 1: Fundamentals of AI and ML	20%	Core AI/ML concepts, techniques, inference types, managed AI services, ML lifecycle, metrics
Domain 2: Fundamentals of GenAI	24%	Tokens, embeddings, context engineering, FMs, limitations, agents, AWS GenAI services, cost tradeoffs
Domain 3: Applications of Foundation Models	28%	FM selection, RAG, vectors, prompting, customization, model evaluation, agent evaluation
Domain 4: Guidelines for Responsible AI	14%	Fairness, bias, robustness, safety, explainability, human review, legal and trust risks
Domain 5: Security, Compliance, and Governance for AI Solutions	14%	IAM, encryption, privacy, prompt security, logging, governance, compliance evidence, data lifecycle

Recommended study priority

Domain 3 first: RAG, prompts, FM selection, evaluation, and customization.
Domain 2 second: GenAI foundations, cost, context engineering, and agent concepts.
Domain 1 third: traditional ML foundations and service selection.
Domains 4 and 5 together: responsible AI, security, compliance, and governance.

3. Start-to-Finish Study Path

Phase 1 : Build the mental model

Study these concepts until you can explain them without notes:

AI vs ML vs deep learning vs GenAI vs agentic AI
Classification vs regression vs clustering
Supervised vs unsupervised vs reinforcement learning
Batch vs real-time vs asynchronous vs serverless inference
Tokens, chunks, embeddings, vectors, context windows
Prompt engineering vs RAG vs fine-tuning vs pre-training vs distillation
Hallucination, nondeterminism, bias, robustness, explainability
Least privilege, encryption, data lineage, logging, retention, residency

Phase 2 : Memorize the service map

Learn services in families rather than isolated definitions:

Text and language: Amazon Comprehend, Amazon Translate, Amazon Lex
Speech: Amazon Transcribe, Amazon Polly
Images and documents: Amazon Rekognition, Amazon Textract
Personalization: Amazon Personalize
ML platform: Amazon SageMaker AI, SageMaker JumpStart
Foundation models and GenAI: Amazon Bedrock, Amazon Nova
Agents and development: Amazon Bedrock AgentCore, Strands Agents, Kiro, Amazon Q, AWS Transform
Business work and analytics: Amazon Quick
Security: IAM, AWS KMS, Amazon Macie, AWS Secrets Manager, AWS PrivateLink
Governance and auditing: AWS CloudTrail, AWS Config, AWS Audit Manager, AWS Artifact, Amazon Inspector, AWS Trusted Advisor

Phase 3 : Learn decision rules

Do not memorize long definitions only. Memorize the condition that causes you to select a service or architecture.

Example:

“Recorded call to text” → Amazon Transcribe
“Text read aloud” → Amazon Polly
“Scanned invoice tables and fields” → Amazon Textract
“Objects in an image” → Amazon Rekognition
“Updated private documents should ground answers” → RAG
“Stable behavior or style must change” → Fine-tuning
“Agent must call tools safely in production” → AgentCore controls + least privilege
“Audit API activity” → AWS CloudTrail
“Retrieve AWS compliance reports” → AWS Artifact

Phase 4 : Practice scenario elimination

For each practice question:

Mark the requirement words: real-time, lowest overhead, regulated, private, frequently changing, human review, audit trail, predict numeric value, extract text from documents.
Eliminate answers solving a different category of problem.
Eliminate unnecessarily complex designs.
Check security, cost, and governance tradeoffs.
Choose the answer that satisfies the requirement directly.

Phase 5 : Final revision

In the final review:

Re-read the service selection tables.
Review the RAG, agentic-AI, and security architecture patterns.
Recite the quick memory rules.
Practice multiple-response, ordering, and matching questions.
Review the April 2026 additions: agentic AI, context engineering, MCP, asynchronous inference, serverless inference, prompt caching, model distillation, LLM-as-a-judge, prompt management, business-alignment metrics, AgentCore security controls, grounding, output validation, and confidence scoring.

4. Core Concepts by Domain

Domain 1: Fundamentals of AI and ML

4.1 AI, ML, deep learning, GenAI, and agentic AI

Concept	Meaning	Typical exam clue
Artificial intelligence (AI)	Broad field of machines performing tasks that usually require human intelligence	Umbrella term
Machine learning (ML)	Subset of AI where systems learn patterns from data	Prediction, classification, forecasting
Deep learning	ML using multi-layer neural networks	Large-scale complex pattern recognition
Generative AI (GenAI)	Models that create new content such as text, images, audio, video, or code	Summarization, drafting, generation, conversational answers
Foundation model (FM)	Large pre-trained model adaptable to multiple tasks	Reuse, prompting, RAG, fine-tuning
Large language model (LLM)	FM focused on language tasks	Text generation, question answering, summarization
Agentic AI	AI that reasons about goals, uses tools, manages steps, and performs workflows	Multi-step execution, tool calls, memory, orchestration

Decision rule

Need a predicted label or number from structured data? Start with traditional ML.
Need generated content, conversational responses, flexible summarization, or language reasoning? Consider an FM.
Need goal-oriented multi-step execution with tools? Consider agentic AI.
Need a guaranteed exact result from fixed rules? Use a rules-based system, not AI.

4.2 Learning types

Learning type	Training signal	Good use case	Common trap
Supervised learning	Labeled examples	Fraud / not fraud, route ticket to team, predict price	Choosing it when no labels exist
Unsupervised learning	Unlabeled data	Customer segmentation, article grouping	Confusing clustering with classification
Reinforcement learning	Rewards and penalties for actions	Sequential decisions, agent behavior improvement	Confusing it with ordinary classification

4.3 Core ML techniques

Technique	Output	Scenario example
Classification	Discrete category	Approve vs review; spam vs legitimate
Regression	Numeric value	Sales forecast; demand amount; price estimate
Clustering	Natural groupings without predefined labels	Segment customers; group similar documents

Fast elimination rule

Output is a category → classification.
Output is a number → regression.
Goal is to discover groups without labels → clustering.

4.4 Inference patterns

Pattern	Best fit	Example	Trap
Batch inference	Large set of predictions on a schedule	Overnight inventory forecast	Paying for real-time when no user is waiting
Real-time inference	Immediate interactive response	Fraud check during payment authorization	Using nightly batch for an online request
Asynchronous inference	Long-running request, caller can retrieve result later	Large document processing	Forcing long workloads into synchronous endpoints
Serverless inference	Intermittent or unpredictable traffic with reduced capacity management	Low-volume prototype or sporadic API traffic	Provisioning an oversized always-on endpoint

4.5 Data concepts

Data concept	Meaning	Examples
Structured data	Defined schema	Relational rows, transaction tables
Unstructured data	No fixed tabular structure	Emails, documents, images, audio
Labeled data	Target outcome is known	Fraud / not fraud records
Unlabeled data	No target label	Raw purchase histories for clustering
Time-series data	Values ordered over time	Demand, sensor readings, sales
Training	Learning or adapting model parameters from data	Model-building stage
Inference	Applying a trained model to new inputs	Prediction or generation stage

lock

4.6 Traditional ML or FM?

This module is part of the full course. Unlock all 73 modules + 65+ practice questions.