AWS AI Practitioner Cheat Sheet (AIF-C01)
Model evaluation metrics, prompt engineering patterns, Amazon Bedrock, SageMaker AI, responsible AI guidelines, and every in-scope AWS service for the AIF-C01 exam
About This Cheat Sheet
A 12-page reference covering every in-scope service from the official AIF-C01 exam guide, organised around the five exam domains: AI & ML Fundamentals (20%), Generative AI Fundamentals (24%), Foundation Models (28%), Responsible AI (14%), and Security & Governance (14%). Covers the full ML-lifecycle vocabulary (supervised vs. unsupervised vs. RL, bias vs. variance, CNN/RNN/Transformer/GAN/diffusion), generative-AI fundamentals (foundation models, embeddings, vector databases, tokenisation), a detailed prompt-engineering taxonomy (foundation / reasoning / structural patterns, inference parameters), Amazon Bedrock (Nova, Titan, Claude, Llama; Knowledge Bases, Agents, Guardrails, custom models), SageMaker AI tools (Studio, JumpStart, Canvas, Data Wrangler, Feature Store, Ground Truth, Clarify, Model Monitor, Pipelines, Model Registry, MLflow), AWS AI services (Comprehend, Rekognition, Textract, Transcribe, Translate, Polly, Lex, Kendra, Personalize, Fraud Detector, A2I), the full Amazon Q family (Developer, Business, in QuickSight, in Connect), model-evaluation metrics (classification, regression, text generation) with decision rules, responsible-AI dimensions with bias types and explainability techniques, and every supporting in-scope AWS service (IAM, KMS, CloudTrail, CloudWatch, Config, Artifact, Audit Manager, Macie, Inspector, Secrets Manager, EC2, ECS, EKS, S3, S3 Glacier, RDS, Aurora, DynamoDB, DocumentDB, Neptune, ElastiCache, MemoryDB, Glue, Glue DataBrew, Lake Formation, EMR, OpenSearch, QuickSight, Redshift, Data Exchange, VPC, CloudFront, Trusted Advisor, Well-Architected Tool, Budgets, Cost Explorer). Built from the official exam guide, the Whizlabs AIF-C01 WhizCard, and a hand-curated personal study sheet — audited against the in-scope list to ensure completeness.
What's Inside
AI & ML Fundamentals
Types of learning (supervised, unsupervised, semi-supervised, self-supervised, RL, transfer learning), structured vs. unstructured data, the bias-variance tradeoff with fixes for under- and overfitting, and the neural-network architectures examiners test: feedforward, CNN, RNN/LSTM, Transformer, GAN, diffusion.
Generative AI Fundamentals
Foundation models vs. LLMs vs. multimodal models, open-source vs. proprietary tradeoffs, embeddings and vector databases (OpenSearch k-NN, Aurora pgvector, Neptune Analytics, DocumentDB, MemoryDB, Bedrock Knowledge Bases), tokenisation and context windows, dimensionality reduction (PCA, SVD, t-SNE/UMAP).
Prompt Engineering & Inference
Customisation ladder from prompt engineering → RAG → fine-tuning → continued pre-training → train-from-scratch with cost tradeoffs. Foundation patterns (zero-shot, few-shot, instruction, role, contextual, style), reasoning (Chain-of-Thought, Self-Consistency, Tree-of-Thought, ReAct, Reflection), structural (output structuring, templates, chaining, RAG). Inference parameters — temperature, top-p, top-k, max tokens, stop sequences, frequency/presence penalty.
Amazon Bedrock
Model families (Amazon Nova, Titan, Anthropic Claude, Meta Llama, Mistral, Cohere, AI21, Stability). Knowledge Bases for managed RAG, Agents for multi-step reasoning + tool calling, Guardrails (content filters, denied topics, PII redaction, grounding checks, prompt-injection detection), custom models via fine-tuning or continued pre-training, Model Evaluation, Provisioned Throughput, Prompt Management.
Amazon SageMaker AI
Studio, JumpStart, Canvas, Data Wrangler, Feature Store, Ground Truth, Training (with Spot), Experiments, Model Registry, Endpoints (real-time, serverless, async, batch), Model Monitor for drift, Model Dashboard, Clarify for bias + SHAP explanations, Pipelines for ML CI/CD, MLflow, Role Manager.
AWS AI Services & Amazon Q Family
Pre-built APIs: Comprehend, Rekognition, Textract, Transcribe, Translate, Polly, Lex, Kendra, Personalize, Fraud Detector, A2I (human-in-the-loop). Amazon Q variants: Q Developer, Q Business, Q in QuickSight, Q in Connect.
Model Evaluation Metrics
Classification (accuracy, precision, recall, F1, AUC-ROC, log loss, confusion matrix), regression (MAE, MSE, RMSE, R²), text generation (ROUGE, BLEU, METEOR, BERTScore, CIDEr, perplexity), with decision rules for when each one applies.
Responsible AI
Seven AWS dimensions (fairness, explainability, robustness, privacy & security, governance, transparency, veracity). Bias types (sampling, measurement, algorithmic, confirmation), detection with SageMaker Clarify, explainability with SHAP and Partial Dependence Plots. Generative AI risks (hallucination, prompt injection, toxic outputs, training-data leakage) with Bedrock Guardrails and RAG mitigations.
Security, Compliance & Governance
IAM least-privilege, KMS encryption, Secrets Manager, Macie for S3 PII discovery, Inspector for CVEs. Audit: CloudTrail, CloudWatch, Config (with custom rules for AI governance), Artifact for compliance reports, Audit Manager for frameworks (HIPAA/GDPR/PCI/AI RMF), Trusted Advisor, Well-Architected Tool with ML Lens and Generative AI Lens.
Supporting AWS Services
Compute (EC2 with GPU / Inferentia / Trainium, ECS, EKS), Storage (S3, S3 Glacier), Databases (RDS, Aurora, DynamoDB, DocumentDB, Neptune, ElastiCache, MemoryDB), Analytics (Glue, Glue DataBrew, Lake Formation, EMR, OpenSearch, QuickSight, Redshift, Data Exchange), Networking (VPC, CloudFront, VPC endpoints).
Cost Management
Cost Explorer, Budgets, Bedrock on-demand vs. Provisioned Throughput, SageMaker Spot training, Serverless Inference, Savings Plans. Free Tier for SageMaker Studio, Bedrock, Comprehend.
Why This Cheat Sheet Helps
The AIF-C01 exam covers a wide range of AI and ML concepts that are easy to confuse under time pressure. Questions frequently ask you to distinguish between similar metrics (ROUGE vs. BLEU vs. METEOR), pick the right prompting technique for a scenario, or identify the cause of model bias. This cheat sheet puts those comparisons side by side so you can internalize the differences at a glance.
It is not a replacement for a full study course — it assumes you already understand the basics. Use it as a final review tool in the days before your exam, or as a quick reference when a practice question exposes a gap in your knowledge.
How to Use It
Skim the cheat sheet from front to back once to get a feel for the topics covered. Then use it as a reference: when you get a practice exam question wrong, find the relevant section and study the tables and examples. On exam day minus three, go through it section by section and make sure every row in every table makes sense to you.
Pair this with CloudNinja's free AI Practitioner practice exam for the best results. The practice exam reveals what you do not know, and the cheat sheet gives you the structured reference to close those gaps quickly.
Frequently Asked Questions
Is this AWS AI Practitioner cheat sheet free?
Yes, this cheat sheet is completely free with no signup required. You can download the PDF directly from CloudNinja and use it as a study reference for the AIF-C01 exam.
Will this cheat sheet alone help me pass the AI Practitioner exam?
No single resource can guarantee you pass. This cheat sheet is a reference for topics you already understand at a basic level — use it alongside video courses, practice exams, and the AWS documentation. It is most valuable as a final review tool in the week before your exam.
Is this cheat sheet updated for the latest AIF-C01 exam?
Yes, this cheat sheet is updated for the current AIF-C01 exam guide. AWS occasionally updates exam content, so always cross-reference with the official AWS exam guide before your test date.
Can I share this cheat sheet with others?
Yes, feel free to share the CloudNinja cheat sheet link with classmates, colleagues, or study groups. We would appreciate if you link back to the CloudNinja page rather than redistributing the PDF file directly.
What topics are on the AWS AI Practitioner (AIF-C01) exam?
The AIF-C01 exam covers five domains: Fundamentals of AI and ML (20%), Fundamentals of Generative AI (24%), Applications of Foundation Models (28%), Guidelines for Responsible AI (14%), and Security, Compliance and Governance for AI Solutions (14%). This cheat sheet covers all five domains with comparison tables and exam tips.
How many questions are on the AIF-C01 exam and what is the passing score?
The AWS AI Practitioner exam has 65 questions (50 scored + 15 unscored) with a 90-minute time limit. You need a scaled score of 700 out of 1,000 to pass. Question types include multiple-choice, multiple-response, and ordering questions.
How should I study for the AWS AI Practitioner certification?
Start with a video course or the AWS Skill Builder to learn the core AI/ML concepts. Then use practice exams to identify your weak areas. Use this cheat sheet as a final review tool in the last few days before your exam — it puts the most frequently tested comparisons and patterns side by side for quick reference.
Keep Studying
Recommended courses
Affiliate links — if you enrol through them, CloudNinja may earn a commission at no extra cost to you.