Digital Garden

AWS Certified AI Practitioner (AIF-C01)

Properties2
tagsaws, aipractitioner
aliasesAIF-C01, AWS AI Practitioner

5 min read

AWS Certified AI Practitioner (AIF-C01)

The AWS Certified AI Practitioner is a foundational certification for people who use, select, or discuss AI and machine-learning solutions on Amazon Web Services without necessarily building the underlying models. The exam emphasizes business use-case selection, core AI and generative-AI concepts, responsible AI, and security/governance decisions.

Exam at a glance

  • Format: 65 questions in 90 minutes; 50 questions are scored and 15 are unscored.
  • Question styles: multiple choice, multiple response, ordering, and matching.
  • Passing score: 700 on a 100-1,000 scale; scoring is compensatory across domains.
  • Target experience: up to six months of exposure to AI/ML technologies on AWS.
  • Out of scope: coding algorithms, implementing data/feature engineering, tuning models, building pipelines, performing mathematical analysis, and implementing security or governance frameworks.

Domain map

  1. AIF-C01 Domain 1 - Fundamentals of AI and ML - 20%
  2. AIF-C01 Domain 2 - Fundamentals of Generative AI - 24%
  3. AIF-C01 Domain 3 - Applications of Foundation Models - 28%
  4. AIF-C01 Domain 4 - Responsible AI - 14%
  5. AIF-C01 Domain 5 - AI Security Compliance and Governance - 14%

Domain 3 is the largest, but Domains 2 and 3 together account for more than half of the scored content. Learn service boundaries and tradeoffs rather than implementation syntax.

Version drift from the supplied PDFs

The local AWS-Certified-AI-Practitioner_Exam-Guide.pdf identifies itself as version 1.4 and the two-page study sheet reflects that older scope. AWS’s live exam guide was revised on April 30, 2026. The live objectives add:

  • agentic AI, multi-agent patterns, tool use, memory, workflow orchestration, and Model Context Protocol (MCP);
  • context engineering and token-cost effects;
  • traditional ML versus foundation-model selection;
  • model distillation, prompt management/versioning, and LLM-as-a-judge;
  • application-level evaluation, task-completion and cost-per-interaction metrics;
  • AgentCore identity/policy, data-leakage prevention, output validation, audit trails, and hallucination grounding.

The current in-scope list adds services such as Amazon Bedrock AgentCore, Amazon Nova, Amazon SageMaker JumpStart, Amazon Quick, Kiro, Strands Agents, and AWS Transform. Amazon MemoryDB, Amazon Fraud Detector, and Amazon Forecast appear in older study material but are not on the current in-scope list. Keep them as historical context, not priority exam targets.

Current in-scope services

Analytics and data preparation

Cost, compute, and containers

Databases and vector-capable stores

Developer and agentic AI tools

  • Kiro - agentic software-development environment.
  • Strands Agents - open-source, model-first SDK for AI agents.
  • Amazon Q - AWS/developer assistance; older Q Business workloads are transitioning toward Amazon Quick.

Machine learning and managed AI

Operations, networking, security, and storage

High-yield service boundaries

  • Prebuilt AI API vs custom ML: choose services such as Transcribe, Translate, Comprehend, Polly, Textract, Rekognition, Lex, or Personalize when the task matches their managed capability; choose SageMaker AI for custom ML lifecycles.
  • Bedrock vs SageMaker AI: Bedrock is the managed generative-AI layer for using and adapting foundation models; SageMaker AI provides deeper control over building, training, hosting, and operating models.
  • RAG vs fine-tuning: RAG supplies retrieved knowledge at inference time and supports citations/fresh data; fine-tuning changes model weights to shape task behavior. Neither guarantees factual output.
  • Artifact vs Audit Manager: Artifact downloads AWS compliance reports and agreements; Audit Manager collects evidence about your own AWS usage against control frameworks.
  • CloudTrail vs CloudWatch vs Config: CloudTrail records API activity, CloudWatch observes metrics/logs/alarms, and Config records resource configuration and evaluates configuration rules.

Sources

Graph View

Backlinks