What Does a Realistic AWS AI Practitioner Study Plan Look Like?
Six weeks is the right target for most candidates approaching AIF-C01 from a non-AWS background. The exam costs $100, has a 14-day waiting period between attempts if you fail, and covers 65 questions across five domains in 90 minutes. A rushed preparation that results in a failed attempt costs more than an extra week of study—both in money and time.
Candidates with existing AWS cloud experience (any Associate or Practitioner level certification) can compress this to four weeks by skipping the service orientation phase. The plan below assumes no prior AWS exposure. Full exam details are at aws.amazon.com/certification/certified-ai-practitioner.
Week 1 — AI and ML Fundamentals (Domain 1, 20%)
Domain 1 covers the conceptual foundation: what distinguishes AI from ML from deep learning, types of learning (supervised, unsupervised, reinforcement), key steps in the ML pipeline (data collection, feature engineering, training, evaluation, deployment), and basic model evaluation metrics (accuracy, precision, recall, F1 score, AUC-ROC).
Primary resource for Week 1: AWS Skill Builder (skillbuilder.aws) — the free tier includes the "AWS Cloud Practitioner Essentials" and "Machine Learning Foundations" courses, which provide the conceptual grounding Domain 1 tests. The paid AWS Skill Builder subscription ($29/month) adds practice question sets, which become essential in Week 5.
Week 1 goal: Understand ML pipeline stages, be able to define and contrast the three main learning types with examples, and know what overfitting, underfitting, training data, and validation data mean without looking them up.
Week 2 — Generative AI Fundamentals (Domain 2, 24%)
Domain 2 covers how generative AI and foundation models work: transformer architecture at a conceptual level, what large language models (LLMs) are and how they generate text, prompt engineering techniques (zero-shot, few-shot, chain-of-thought), retrieval-augmented generation (RAG), and fine-tuning.
Key concepts to master in Week 2:
- Foundation models vs. traditional ML models — Foundation models are pre-trained on large datasets and adapted to specific tasks; traditional models are trained from scratch for each task.
- Prompt engineering techniques — Same as Google AI Essentials but tested in the context of AWS Bedrock: zero-shot, one-shot, few-shot, chain-of-thought, and system prompts.
- RAG architecture — Combines a retrieval system (like Amazon Kendra) with a generative model to ground outputs in specific, up-to-date documents. Frequently tested in Domain 3 scenario questions.
- Fine-tuning vs. RAG — When to use each: fine-tuning for adapting model behavior; RAG for grounding responses in specific knowledge bases without retraining.
Week 3 — AWS AI Services Deep Dive (Domain 3, 28%)
Domain 3 is the largest and most AWS-specific. This week requires learning what each AWS AI service does, what use cases it solves, and when to choose it over alternatives. The services to know:
- Amazon Bedrock — Managed access to foundation models. Know the model providers available (Anthropic, AI21 Labs, Cohere, Meta, Stability AI), how to access models via API, and the Agents and Knowledge Bases features.
- Amazon SageMaker — Full ML platform. For AIF-C01 purposes, know SageMaker at the use-case level: training custom models, hosting endpoints, SageMaker Studio for experimentation, SageMaker Autopilot for automated ML.
- Amazon Rekognition — Image and video analysis: object detection, facial analysis, text recognition, content moderation.
- Amazon Comprehend — NLP: entity recognition, sentiment analysis, topic modeling, custom classifiers.
- Amazon Transcribe — Speech-to-text with speaker diarization and custom vocabulary.
- Amazon Translate — Neural machine translation.
- Amazon Polly — Text-to-speech.
- Amazon Kendra — Enterprise search; primary retrieval component in RAG architectures tested on the exam.
- Amazon Textract — Document text and data extraction from scanned PDFs and images.
Best resource for Week 3: AWS service documentation FAQs (each service's documentation page) and the official AIF-C01 exam guide, which lists all services tested.