AWS Certified Generative AI Developer Professional Exam Prep

Explore the AWS certification exam overview, study methods, and fundamentals of AI and machine learning, plus an introduction to AWS services with practice questions.

Explore the AWS learning path across foundational, associate, and professional levels, comparing video courses and official docs while tailoring study style to individual needs, with practice questions and hands-on exercises.

Explore the AWS certification exam landscape, with foundational, associate, and professional levels, no prerequisites for the professional track, and the exam's duration and content domains.

Discover AWS management tools—the management console, CLI, and SDK—and learn infrastructure as code with CDK and CloudFormation, plus IDE tooling like the AWS toolkit.

Create a new AWS account and set up an IAM user to manage permissions with administrator access, sign in as IAM user rather than root, and enable MFA.

Define ai and explain machine learning, deep learning, and generative ai using features and label to predict outcomes, while briefly tracing ai history and cloud-based data analysis.

Compare traditional programming with explicit rules to machine learning that learns patterns from data, builds models, and predicts unseen data through training, testing, and evaluation.

Deep learning uses a multi-layered neural network modeled after the brain, with big data. It automatically extracts features and learns complex patterns for image recognition and speech recognition.

Explore how deep learning advances in natural language processing gave rise to generative ai and large language models, built on transformer architecture with attention and word embeddings.

Learn supervised and unsupervised learning by modeling input-output relationships with labeled data and error-driven training. Explore classification, regression, clustering, using examples like ice cream sales vs temperature and grouping data.

Explore unsupervised learning, where the model finds inherent structure in unlabeled data without a teacher, using clustering to group similar data and reveal natural patterns.

Explore reinforcement learning, where an agent learns optimal actions through trial and error by receiving rewards or penalties, to maximize long-term performance, as seen in gaming and autonomous driving.

Explore data types and characteristics, including structured, semi-structured, and unstructured data with examples like csv, json, avro, parquet, and orc. Compare batch processing and streaming for timing data collection.

Explore real-world AI applications across computer vision, including image classification and object detection, and natural language processing, using AWS services like Rekognition, Comprehend, Translate, Forecast, and Fraud Detector.

Explore the end-to-end ml pipeline—from data collection, preprocessing and analysis, training, evaluation, operation, deployment, and inference—highlighting key considerations for each stage.

Begin data collection at the start of the ml pipeline with high-quality, relevant data. Define the problem, identify data and sources, ensure relevance, volume, accuracy, fairness, and cost constraints.

learn how to train ai models using prepared data, tune hyperparameters, avoid overfitting with train-test splits, and optimize performance with gradient descent and hyperparameter tuning.

Evaluate a trained model on unseen test data to detect overfitting, using the confusion matrix and metrics like accuracy, precision, recall, F1, ROC AUC, plus regression metrics and cross-validation.

Deploy trained models to production with scalable endpoints and monitoring, then perform real-time or batch inference on unseen data to generate business-valued predictions via APIs.

Explore the technical evaluation and social responsibility of AI, focusing on data bias, fairness across groups, and GDPR-related risks, using tools like Amazon SageMaker Clarify.

Explore MLOps, the DevOps for machine learning, automating from development to production with CI, CD, continuous training, drift monitoring, and SageMaker experiments.

Compare custom and pre-trained models and apply transfer learning to adapt general knowledge to your tasks. Leverage AWS AI services to access pre-trained models via API with fine-tuning options.

Generative artificial intelligence creates new original content—text, images, audio, or code—by learning patterns from massive data and applying them to novel outputs.

Leverage foundation models trained on internet-scale data that use self-supervised learning to predict masked text from unlabeled data, and apply transfer learning to diverse tasks.

Foundational models vary in performance and purpose; parameter counts measure scale, with larger models offering higher performance but operational cost, and the context window acts as short-term memory affecting latency.

Explore foundational models that handle multiple modalities, from text to images and audio, and learn how diffusion models generate high-quality images from noise using prompts.

Explore how generative AI processes language with tokens and tokenization, embeddings, and chunking, and how overlapping chunks enable semantic search through vectorized meaning.

Explore prompt engineering to maximize foundation model outputs by crafting clear, specific instructions, using zero-shot, few-shot, and chain-of-thought prompts with best practices for context, format, and iterative refinement.

Explore how inference controls the AI’s responses, using temperature, top p, and top k to balance creativity and consistency, and apply stop sequences to manage output and nondeterminism.

Tailor general foundation models with internal data using customization and retrieval-augmented generation. Prepare data by chunking and embedding it into a vector database, then augment prompts with retrieved information.

An agent uses a foundational model to autonomously plan, choose tools, execute actions, log results, and fine-tune for autonomous problem solving in repeating cycles to achieve goals.

Assess the risks of generative AI, including hallucination, interpretability, bias and fairness, copyright, and misuse, and apply safeguards and factual information to ensure safe, responsible use.

Identify and explain the main prompt attacks in generative AI, including prompt injection, Jane breaking, prompt rigging or exposure, prompt poisoning, and hijacking, and their impact on safety.

Assess generative AI using benchmarks such as glue/superglue, emmylou, and the hugging face leaderboard, plus automatic metrics like rouge and bleu, then supplement with human evaluation.

Explore how AWS layers AI services from managed APIs to generative AI platforms, usable without AI expertise. Discover Amazon Bedrock, Amazon SageMaker, and infrastructure for language, speech, and image processing.

Leverage Amazon Polly to generate real-time speech from text, with 50 voices in 25 languages, and refine output using SSML and custom lexicons.

Explore Amazon Polly text-to-speech in the AWS console, compare neural and standard engines, select English Joanna voice, use SSML for emphasis and pauses, and save synthesized speech to S3.

Learn how Amazon Transcribe converts audio to text or from uploaded files; it supports 16 languages, identifies 2 to 10 speakers, timestamps, punctuation, a custom vocabulary, and masks sensitive data.

perform hands-on with Amazon Transcribe to enable real-time speech recognition, stream audio, select English, export transcripts, and apply custom vocabulary, filtering, and PII masking.

Explore Amazon Translate, a high quality automatic translation service powered by natural language understanding and deep learning that supports real-time translation, back translation, customer terminology, and custom translation models.

Explore Amazon Translate in real-time and large-document modes, using auto-detect to translate English to Spanish, and apply custom terminology with csv, tmx, or tsv files and parallel data from S3.

Explore Amazon Comprehend, an advanced natural language processing tool that reads longer texts, detects language, and extracts phrases. Analyze sentiment and recognize entities, and you can train a custom model.

Explore Amazon Comprehend's real-time text analysis to extract entities, keywords, language, sentiment, and PII, with batch analysis for large documents stored in S3.

Discover amazon lex for building chatbots and voice bots with conversational ai, using alexa technology, integrated with aws lambda and amazon cognito for business workflows like booking rooms and private skill.

Amazon Kendra enables high performance semantic search across data sources like S3, SharePoint, and OneDrive, with secure per-user results and log-pipeline knowledge-base integration for Bedrock.

Explore Amazon Personalize by building a simple dataset from purchase history and viewing history, then create a user and generate tailored movie recommendations using the magic movie machine demo.

Explore Amazon Rekognition's deep learning analysis of images and videos, detecting objects and faces, estimating gender, age, and emotion, enabling inappropriate content detection and training custom models from your images.

Explore Amazon Rekognition's image and video analysis in real time, including label detection, facial analysis, age and gender, face comparison, text detection, celebrity recognition, PPE detection, and video tracking.

Explore Amazon Textract, an ocr service that extracts text from pdfs and images and understands tables and forms. Retrieves id details such as name, date of birth, and address.

Explore Amazon Textract in a hands-on demo: perform high-performance OCR on handwritten and printed documents, extract text, analyze forms, tables, signatures, receipts, and IDs, and run queries to answer questions.

Amazon Q is a fully managed generative AI assistant for business chat, enabling developer coding help, VS Code, QuickSight visualizations, and Connect and AWS console integrations.

Explore Amazon Bedrock, a fully managed service that unifies foundation models via a single aws api, with data ingestion, vectorization, agents, and guardrails for safe generative ai.

Explore Amazon Bedrock Playground with hands-on text and image generation using Titan Text G1. Compare models from the model catalog and adjust prompts; learn pay as you go, serverless Bedrock.

Do a hands-on creation and testing of an Amazon Bedrock knowledge base from the AWS whitepaper, using S3 data, Titan embeddings, and OpenSearch for retrieval-augmented generation.

Explore Amazon SageMaker as an integrated AI platform that builds and trains machine learning models from custom to pre-trained, guiding data collection, processing, training, evaluation and monitoring across the pipeline.

Configure an Amazon SageMaker domain using quick single-user setup or organization setup, selecting authentication via Identity Center or user login, and tailor roles, applications, and networking.

Learn image classification with AWS SageMaker Jumpstart by deploying a pre-built TensorFlow MobileNet model, testing via Studio and a Jupyter notebook.

Use Amazon SageMaker Data Wrangler on SageMaker canvas, import data from S3, Snowflake, and Databricks, perform GUI-based transformations with no coding, run on EC2, and save to S3 or pipelines.

Explore data transformation with SageMaker Data Wrangler on canvas, loading a csv, applying no-code transformations like fill missing values and one-hot encoding, and exporting results.

Explore Amazon SageMaker feature store, a central repository for managing ML features that ensures training and inference data consistency and prevents drift, with online low-latency and offline batch stores.

Create a SageMaker feature group in the feature store with online and offline options, configure an S3 offline bucket and IAM role, and ingest data via Jupyter Lab.

Discover supervised datasets for training machine learning models with SageMaker Ground Truth, a managed service enabling labeling via Mechanical Turk or private workforce for image recognition and natural language processing.

Learn how to deploy Amazon SageMaker endpoints, including real-time serverless, real-time inference, asynchronous inference, and batch transform, with multi-model endpoints, multi-container endpoints, and serial inference pipelines.

Train a SageMaker model on the iris dataset using XGBoost to classify three iris species with softmax, using 100 training rounds and data prepared in S3.

Deploy a trained iris model in SageMaker Studio using XGBoost, create an endpoint, and run a prediction to verify inference.

Conduct a hands-on evaluation of a deployed iris classifier using iris dataset in SageMaker Studio, computing accuracy, confusion matrix, and classification report on a 30-sample validation set, achieving 100% accuracy.

Explore SageMaker inference recommender to run load tests and select optimal machine learning inference instance types by latency and throughput, and compare with AWS Compute Optimizer's CloudWatch-based recommendations.

Orchestrate end-to-end ML workflows with SageMaker pipelines, a serverless dag-based platform for data prep, training, monitoring, and ML lineage tracking, with scheduling via EventBridge.

Monitor automatically watches production data and models, establishes a baseline, detects drift, and triggers alerts or retraining while evaluating data quality, accuracy, bias, fairness, and feature attribution.

Discover how the SageMaker model registry centralizes management, versioning, and tracking of trained ML models and their groups, with artifacts, metadata, bias analytics, evaluation data, and hyperparameters.

Discover the SageMaker studio model registry: create a model group and collection, register a Jumpstart image classification model (mobilenet v2 1.0), and manage training and deletion permissions.

Leverage Amazon SageMaker Clarify to detect bias and explain model decisions, assess fairness and transparency, and evaluate foundation models on text summarization and Q&A with rouge and bot score.

Explore the SageMaker role manager, which creates SageMaker IAM roles via preset personas. Data scientist gets SageMaker studio and learning pipeline permissions; MLOps manages pipeline tasks and S3 access.

Explore how the SageMaker role manager creates IAM roles from user personas. Select activities, configure policies and tags, and apply the SageMaker prefix.

Discover how AWS Organizations centralizes governance across accounts with organization units and service control policies. Compare deny list and allow list strategies, and manage consolidated billing and quotas across organization.

Learn how AWS organizations use CloudFormation stack sets to deploy across accounts and regions. Centralize logs with organization trails and delegate security operations to member accounts for findings.

Explore AWS organizations by creating an organizational unit, attaching a service control policy that restricts EC2 instance creation to t2.micro, and validating inheritance.

Create and govern approved AWS resources using the AWS Service Catalog by building products from CloudFormation templates, organizing portfolios, applying constraints, and enabling end users to launch compliant resources.

Explore system monitoring across infrastructure and services, including servers, networks, operating systems, DNS, and load balancers, with audit trails and change history, using AWS CloudWatch and AWS CloudTrail.

Monitor AWS resources with CloudWatch by collecting logs and metrics, triggering alarms, and sending SNS notifications; query logs with CloudWatch Logs Insights and analyze via S3, Athena, or QuickSight.

CloudWatch collects metrics from AWS services and on premises systems via the CloudWatch agent, including custom metrics, to monitor performance; it uses alarms, EventBridge rules, and SNS or Lambda actions.

Install, configure, and start the CloudWatch agent via System Manager, attach the CloudWatch agent Admin policy and CloudWatch Agent Server policy with least privilege, and configure metrics and logs.

Learn to monitor an EC2 instance with CloudWatch, send metrics via CloudShell, and configure an alarm that stops the instance when thresholds are exceeded, plus explore CloudWatch logs and metrics.

Create an IAM role for Lambda and build a Python 3.11 function that outputs EMF formatted logs to CloudWatch metrics. Deploy, test, and view EMF-based metrics by namespace and dimensions.

Explore how AWS CloudTrail records user activity and API calls as trails, logging management and data events, with S3 delivery, encryption, integrity checks, and optional insights for unusual activity.

Learn the basics of aws cloudtrail, create and manage trails, view event history for 90 days, store logs in s3, and integrate with cloudwatch logs and iam roles.

AWS config logs resource changes and evaluates compliance with organizational rules, tracks change history and timings, enables centralized governance with multi-account aggregation, and integrates with EventBridge, SSM, and conformance packs.

Learn hands-on how AWS Config detects non-compliant EC2 instances by type and automatically stops them via a remediation action using an IAM role.

Discover AWS X-Ray, a distributed tracing service that monitors complex microservices, visualizes service maps and traces, and helps identify bottlenecks, latency, and errors across requests.

Enable aws x-ray traces for a lambda function, view the service map and traces in CloudWatch, and analyze latency to identify bottlenecks.

Explore AWS Systems Manager as a hybrid environment operational tool, covering patch management with a patch baseline, maintenance windows, run command, inventory, and compliance reporting across managed instances.

Master AWS Systems Manager by exploring node, change, operation, and application management, and using tools like Session Manager, Run Command, and State Manager to automate and secure environments.

Learn to manage nodes with AWS Systems Manager using automation, run command, and patch manager, including creating an IAM role, launching EC2 instances, and applying patch policy.

CloudFormation enables infrastructure as code by using YAML or JSON templates to define AWS resources such as EC2 instances and networks, then reproduce environments for dev, test, and production.

Learn how AWS CloudFormation templates define resources through sections like parameters, mappings, conditions, transform, and outputs, and deploy across accounts using stacks and stack sets.

Preview changes before deployment with AWS CloudFormation change sets, showing what is added, removed, or modified. Create, review, and execute the change set, or delete it if issues arise.

Identify drift between CloudFormation templates and actual resources, and integrate drift detection with AWS Config to auto notify and remediate when drift occurs.

Review how CloudFormation rollback restores a stack to its last stable state after deployment failures, using retry, update, or rollback with CloudWatch triggers.

Learn how CloudFormation resource attributes control resource interactions, lifecycle policies, and update behavior, including creation, deletion, depends_on, and rolling update policies for autoscaling groups and lambda aliases.

Learn to deploy ec2 instances using CloudFormation by authoring yaml templates, creating stacks, and applying change sets with parameters, ami, and instance types in us east 1.

Explore hands-on CloudFormation features like rollback, drift detection, change sets, and stack policies while creating and validating an S3 bucket with Infrastructure Composer, templates, and policy control.

Leverage the AWS CDK to define infrastructure with TypeScript and Python, generating CloudFormation templates and deploying via stacks and constructs, using scope and id for organization and grants for IAM.

Learn how the CDK pipeline defines the entire CI/CD process with CodePipeline and CloudFormation, including bootstrapping, synthesis, nested stacks, self-mutation, and deploying updates from GitHub.

Build an AWS CDK pipeline that automates CI/CD, creating an S3 bucket on git push to GitHub, using Cloud Shell and Secrets Manager.

Learn how AWS identity and access management controls fine-grained permissions across resources and accounts, using users, groups, roles, and policies to enforce least privilege.

Define and manage AWS IAM policies by using identity-based and resource-based policies, inline and managed policies, written in JSON with version 2012-10-17, including conditions, principals, and resources.

Explore RBAC and ABAC using AWS IAM policies to grant billing access and control RDS actions via tags such as environment production.

Set up a cross-account IAM role with an external ID on the trust policy, grant Amazon S3 read-only access, and test with STS assume-role to verify its behavior.

Explore cloud data encryption for data at rest and in transit, including symmetric and asymmetric methods, hashing, and hybrid approaches, with AWS standards like FIPS.

Learn how AWS KMS creates and manages cryptographic keys, including CMKs and AWS managed keys, with bring your own key, automatic rotation, and the key lifecycle from creation to deletion.

Perform the AWS KMS data keys hands-on flow, generating a data key, encrypting with OpenSSL, decrypting the data, and managing a customer managed key with a seven-day deletion wait.

Explore AWS Secrets Manager, a centralized service for securely storing and rotating secrets with KMS. Learn access control, automatic rotation via Lambda, and its differences from Systems Manager Parameter Store.

Create and configure secrets in AWS Secrets Manager using key-value pairs, optionally rotate with Lambda, and retrieve, update, or delete secrets via Cloud Shell commands.

Map AWS security functions identify, protect, detect, respond, and recover to services such as Organizations SCP, System Manager, AWS Config, Amazon Detective, Amazon Inspector, Amazon GuardDuty, Shield, Lambda, and CloudFormation.

Amazon Macie uses machine learning and pattern matching to discover and classify sensitive data in S3, using managed and custom data identifiers to generate severity-based findings.

Explore how Amazon Macie detects PII in S3 buckets, configure custom identifiers for employee IDs, run one-time or scheduled discovery jobs, and view findings to safeguard sensitive data.

Explore how Amazon Inspector automates vulnerability management, detects unintended network exposure, and scans EC2 instances, container images in ECR, and Lambda functions against CIS benchmarks and CVSS scoring.

Explore how AWS Audit Manager automates data collection, continuous monitoring, assessment, and detailed reporting to streamline regulatory and internal compliance audits across AWS resources.

Compare object, block, and file storage, noting object storage uses blobs with metadata and http/https access for scalable archiving and web content, with aws s3 options.

Explore Amazon S3's durable object storage with buckets, diverse storage classes including standard and glacier, and features like versioning, encryption, and lifecycle management for data lakes and static hosting.

Explore how S3 secures data with access control via IAM policies, bucket policies, and ACLs, enables signed URLs, server-side or client-side encryption with KMS, versioning, and object locking.

Master S3 operations from Getobject and list bucket to batch operations, multipart upload, transfer acceleration, and S3 select, then explore data lake concepts with Athena, Glue, Redshift, and EMR.

Learn S3 basic operations with hands-on demos by creating a bucket, uploading a sample HTML file as a static website, configuring lifecycle rules, and generating presigned URLs for temporary access.

Configure S3 cross-origin requests by creating a resource and a website bucket, applying policies, enabling static website hosting, and updating CORS to let the website fetch the resource bucket's images.

Upload a text file to the S3 input folder; a bucket event triggers a Lambda function to uppercase the text and save the result to the output folder.

Learn how EBS and EFS provide persistent storage for EC2: EFS offers cross-AZ shared file storage, while EBS is single-AZ block storage, with encryption and snapshots.

Explore creating and mounting an Amazon EFS file system to an EC2 instance, configure security groups for NFS, and customize throughput modes, backups, and encryption settings.

Encrypt unencrypted EBS volumes and EFS file systems by creating encrypted snapshots, copying with the default KMS key, and restoring as new encrypted resources.

Discover how the AWS storage gateway links on-prem storage with AWS, enabling data synchronization for backup and disaster recovery via four gateway types, including file, volume, tape gateways.

Explore Amazon EC2 virtual servers, high availability across multiple availability zones, and pay-as-you-go pricing. Learn about instance types, AMI templates, and placement groups for scalable, fault-tolerant computing.

Explore how aws ec2 instance types classify workloads across general-purpose, compute, memory, storage, and accelerated categories, including t series bursts and dedicated host or dedicated instance options.

Explore EC2 purchasing options, including reserved instances with long-term discounts, on-demand pricing, saving plans, and flexible spot instances; learn when to use capacity reservations and how interruptions affect workloads.

Master AWS auto scaling concepts, including EC2 and application scaling with target tracking, step, simple, scheduled, and predictive policies, plus cool down, warm up, health checks, and lifecycle hooks.

Define a container as a lightweight, standalone package with code, runtime, system, and settings. Explain its isolation from environments and outline AWS container services including ECS, EKS, EC2, and Fargate.

Explore how Amazon ECS, a fully managed container orchestration service, runs Docker containers on AWS and separates the control plane from the data plane to manage clusters, services, and tasks.

Launch an EC2-based ECS cluster, configure an auto scaling group with Amazon Linux on a t2 micro, deploy a task running an http server container, and use CloudFormation.

Learn how to create and run ECS clusters on Fargate, define tasks with nginx, configure networking and metrics, and enable capacity providers, Fargate spot, scheduling, and EC2 alternatives.

Create an EKS cluster and EC2-based node group by configuring IAM roles, selecting version 1.33, enabling public and private API access, and adding the VPC CNI, CoreDNS, and kube proxy.

Explore AWS Lambda, a serverless, event-driven compute service that runs code in response to triggers like HTTP requests, S3 uploads, or schedules, with examples using API Gateway and DynamoDB.

Discover the AWS Lambda execution environment, its init, invoke, and shutdown lifecycle, cold and warm starts, and how provisioned and reserved concurrency manage latency and throttling.

Explore how AWS Lambda uses event source mapping to batch items from SQS, Kinesis, and DynamoDB streams, providing at least once delivery, with optional filtering and asynchronous invocation.

Create a new AWS Lambda function from scratch with Node.js 20 runtime and an IAM execution role, then test with a sample event and explore versions and aliases.

Link the api gateway to a lambda function using version aliases to separate development and production, and implement a canary release with weighted routing for the new feature.

Learn how AWS batch runs hundreds of thousands of containerized jobs in parallel for large-scale batch processing, choosing Fargate or EC2 and saving results to S3.

Summarize AWS network services using a typical diagram, detailing a VPC with public and private subnets, internet gateway and NAT gateway, VPN connections, and security groups for high availability.

Explore Amazon VPC concepts, including private networks, cidr blocks, subnets across availability zones, public versus private subnets, internet gateway, nat gateway, and elastic ip management.

Explore how security groups, as stateful, control instance level traffic with allowlists, and how subnet level network ACLs add a stateless defense in depth, including ephemeral port considerations.

Configure a multi-AZ VPC from scratch in us-east-1, create private and public subnets, set up security groups and ACLs, and provision an Internet gateway and NAT gateways for EC2 connectivity.

Configure a vpc with private and public subnets, attach an internet gateway, and set up the load table to route traffic between subnets and the internet.

This hands-on session guides you through creating security groups for public and private subnets and configuring a network ACL, with SSH inbound rules and subnet-level firewall behavior.

Discover how VPC endpoints provide private access inside a VPC, comparing interface and gateway endpoints for S3 and DynamoDB, with PrivateLink, DNS, and endpoint policies to control access.

Demonstrates creating and testing a VPC endpoint for S3 with EC2 and IAM role, including gateway and interface endpoints and session manager checks.

Explore gateway options connecting your VPC to the internet, including internet gateways, NAT gateways, and NAT instances, with emphasis on IPv4/IPv6 translation, egress-only gateways, inbound/outbound rules, and high availability.

Learn how a content delivery network caches content at AWS edge locations to offload the origin, reduce latency, and enable DDoS protection with CloudFront.

Discover how CloudFront secures content with default ssl, ddos protection via waf and shield, geolocation and ip-based distribution, origin access controls using oai and oac, plus signed url and cookies.

Explore CloudFront edge functions that run at edge locations to customize requests and responses, enable basic authentication and authorization, and reduce origin load with CloudFront functions and Lambda@Edge in Node.js/Python.

Explore setting up CloudFront with origin access control to block direct S3 access and deliver content exclusively through CloudFront, including creating a private S3 bucket and validating access via distribution.

Explore hands-on use of CloudFront functions to add a response header. Use Lambda@Edge to rewrite URLs for an S3 origin, highlighting lightweight versus heavier processing.

Explore AWS Direct Connect, offering dedicated and hosted connections with private, public, and transit virtual interfaces, Direct Connect gateway and transit gateway to connect VPCs across regions, with VPN backup.