AWS Certified AI Practitioner (AIF-C01)

Most organizations are adopting AI faster than their teams can evaluate it. When analysts, managers, and cloud staff can't tell a foundation model from a managed service, or a guardrail from a governance framework, AI projects stall or carry unmanaged risk.

Role fit: This beginner course suits IT Practitioners, Team Leads, analysts, and managers who will guide AI initiatives without building models from scratch. It is a good first step before deeper AWS machine learning or generative AI developer training.

Time investment: About 16 hours across 19 skills, typically 45–60 minutes each, so learners can complete one skill per session.

Risk and compliance: Later skills cover responsible AI governance, generative AI security, hallucinations, bias, and related data considerations. This gives your team a shared vocabulary for audit and compliance conversations.

Change-management note: The course includes AWS account readiness, including root account MFA, an IAM account, and billing alerts. For services such as Amazon SageMaker AI and other managed AI tools, learners are generally guided to watch the demos rather than provision resources. If a Team Lead or Training Manager wants learners to experiment outside the course, use a governed sandbox account.

End-of-skill challenges help learners find knowledge gaps before the AIF-C01 exam. Team Leads can use Team Reporting to track progress across the team.

IT teams complete this training to move AI conversations from hype to grounded decisions. Instead of guessing which AWS service fits a business request, your team learns to match real scenarios to the right tool and flag risk before a pilot reaches production.

• Build internal knowledge assistants with guardrails. Learners work hands-on in Amazon Bedrock to stand up a retrieval augmented generation (RAG) app backed by a knowledge base. The course frames it around use cases like answering employee questions from HR policies and onboarding docs, or helping engineers triage issues from runbooks and logs. They then apply guardrails to limit hallucinations and off-topic use.
• Choose the right managed AI service the first time. A scenario-based scavenger hunt has learners map business cases to AWS managed AI services such as Amazon Comprehend, Amazon Translate, Amazon Transcribe Medical, and Amazon Kendra. That means fewer false starts and less reliance on outside help for early scoping.
• Control cost and governance from day one. Learners compare foundation models on capability, performance, cost, and compliance needs. They also examine token-based cost drivers and practice setting billing alerts so experiments don’t create surprise spend.
• Speak the language of responsible AI. The course covers fairness evaluation concepts, SageMaker AI Clarify, shared responsibility, and the machine learning lifecycle from business objectives through technical tasks. Team Leads and Training Managers can rely on staff who understand responsible AI, security, and compliance risk well enough to partner with technical teams and stakeholders.

Knowledge

• How AI, machine learning, deep learning, and generative AI differ, including supervised, unsupervised, and reinforcement learning and where each fits
• ML workflow phases and performance metrics such as area under the ROC curve, plus trade-offs between edge-deployed small models and cloud-hosted large language models
• How to select a foundation model in Amazon Bedrock by task type, capability, performance, operational requirements, and compliance needs
• Retrieval augmented generation (RAG) concepts, including knowledge bases, vectors, tokens, and how token usage affects application consumption and billing
• Which AWS managed AI services fit which business problems, including Amazon Comprehend, Translate, Textract, Rekognition, Kendra, Lex, Polly, Transcribe, and Personalize, plus AWS AI infrastructure such as Trainium and Inferentia
• Responsible AI and governance principles: model cards, data lineage, bias and quality checks in SageMaker AI Pipelines, and generative AI risks such as hallucinations and non-deterministic outputs

Ability

• Enable model access and test prompts in the Amazon Bedrock playground, including adjusting temperature
• Build a RAG application on a Bedrock knowledge base for use cases like internal policy assistants or troubleshooting runbooks
• Configure guardrails to keep AI applications on topic and reduce inappropriate or fabricated responses
• Refine prompts to improve output clarity, relevance, and consistency for common generative AI tasks
• Match real-world scenarios to the right AWS managed AI service and set billing alerts to manage experimentation
• Communicate AI governance, compliance, and fairness requirements to technical teams and align training with AIF-C01 exam objectives