Complete AWS Certified AI Practitioner – AIF-C01 [NEW]

Amazon Showbusiness, a no-code AI-powered assistant, connects to data sources, generates content, and automates tasks with plugins via a simple web interface, featuring guardrails and access control.

Set up an Amazon Q business application connected to a large language model and data sources to create a searchable workplace knowledge base and tools like a sales script generator.

Assign and test new users with sign-in and MFA for the deployed Amazon Q business application, then connect a data source and enable responses from approved sources with Pro.

Set up a knowledge base for the Amazon Q business application using a native retriever and S3 data sources, including a CSV employee file. Test salary queries against approved sources.

Configure global controls to manage how the data source and the large language model are used, toggling approved sources, LM fallback, topic controls, and blocked words.

Explore the blocked words feature by adding salary-related terms like 'earn' and observe how the application prevents them from appearing, while also examining topic-specific controls.

Explore topic specific controls in a conversational ai system. Create topics with natural language descriptions, set blocking rules, and test how salary or name queries are handled.

Hands-on: creating Amazon Q apps demonstrates building an app to auto-generate employee success stories from a knowledge base, using prompts, inputs, and library publishing.

Unsubscribe the user and delete the application, then remove the user from IAM Identity Center to clean up resources. Follow best practices by deleting resources after completion.

Amazon Transcribe converts audio to text in multiple languages and accents for subtitles and transcripts, enabling batch and real-time HIPAA-compliant transcription with custom vocabulary, speaker identification, and auto punctuation.

Explore amazon transcribe to convert audio or video to text, manage s3 buckets, run real-time transcription, set languages and content filtering, and download the json transcript.

Explore Amazon Polly, Amazon's text-to-speech service, delivering natural sounding voices across languages with real-time synthesis, custom lexicons, and speech marks for media, education, and customer service use cases.

Explore Amazon Polly pricing and models, from per-character pay-as-you-go with a 12-month free tier, to standard, neural, long-form, and generative TTS engines for expressive voices.

Explore Amazon Polly by testing different text-to-speech engines, comparing generative and long form voices, and saving outputs to an S3 bucket for practical evaluation.

Explore how to customize speech output with SSML in Amazon Polly, using pauses, emphasis, phonetic pronunciations, whispering, and language-tagged voices to tailor tone and pace.

Discover real-time and batch translation with Amazon Translate, featuring custom terminology, brevity, and formality controls, auto-detected sources, and document translations for PowerPoint and text files into Italian and Bulgarian.

Explore batch translation with Amazon Translate by creating a batch job in S3, choosing a file type, source and target languages, and a custom terminology file.

Explore diffusion models as a key generative AI approach, learn forward and backward diffusion and the denoising training framework to produce high-quality images, audio, and text, with super-resolution and inpainting.

Large language models are trained on data to understand and generate text using tokens, embeddings, and vectors. They can be multimodal and handle questions, summarize, and describe images.

Explore generative adversarial networks and variational autoencoders, comparing two-network adversarial training with generator and discriminator to encoder–decoder architectures mapping data to latent distributions for image, text, and data generation.

Explore the benefits and practical use cases of generative AI, including versatility, real-time content generation, automation, personalized outputs, scalability, and applications across media, retail, healthcare, finance, and manufacturing.

Explore the challenges of generative AI, including compliance and privacy guardrails, anonymization techniques, audits, social risks, filtering, content moderation, verification of generated content, data protection, security, and interpretability.

Identify your task and compare generative AI models by accuracy, reliability, data needs, cost, and compliance to select a fit, considering Titan, Jurassic two, Claude AI, and Stable Diffusion.

Learn five key metrics to measure the business impact of generative AI—user satisfaction, ARPU, cross-domain performance, conversion rate, and efficiency, using surveys and net promoter score.

Explore the generative AI application lifecycle from ideation to deployment, including use-case identification, feasibility, model selection, fine-tuning, evaluation, ethics, and ongoing maintenance.

Explore Amazon Bedrock, a serverless platform for generative AI that provides diverse foundation models, fine-tuning with knowledge bases, and agents for task automation, with secure, region-bound data handling.

Explore a sample Amazon Bedrock architecture that automates e-commerce product review and description generation using Amazon Rekognition for image analysis, Lambda orchestration, API Gateway, and Bedrock Titan for tailor-made descriptions.

Explore practical use cases of Amazon Bedrock, including content creation, customer support chatbots, data augmentation, and personalized product recommendations on a secure, scalable, fully managed platform.

Explore Amazon Bedrock to access large language models for text and image generation, manage model access, test prompts in the playground, and begin building applications.

Install Visual Studio Code by downloading the Windows installer, run the installation wizard, accept the agreement, and keep default settings, then finish to launch the app.

Set up Visual Studio Code to invoke a model from Python, connect to AWS by creating an IAM user, install boto3 and AWS CLI, and configure credentials and region us-east-1.

Invoke the Amazon Titan model in Bedrock via Boto3, sending a meeting transcript as the prompt and extracting the action items from the response.

Explore hands-on image generation in AWS bedrock by using the Titan image generator v1 with text prompts, saving outputs to a local folder, and iterating prompts and configurations.

Explore why improving foundation models matters, from accuracy and relevancy through domain-specific fine-tuning to policy-aligned outputs. Enhance efficiency, scalability, and compliance while aligning models with business goals.

Explore retrieval augmented generation, or rag, which enhances large language models by querying an external knowledge base with vector embeddings and fast similarity search to deliver up-to-date, trusted responses.

Set up a knowledge base in Bedrock by uploading data to S3, creating a vector store with OpenSearch, and testing with a model to empower agents.

Leverage foundational models through autonomous agents that observe, decide, and act to achieve defined goals. They integrate with databases and CRMs, perform tasks, and gather data to refine models.

Set up an agent to compute currency exchange rates using a knowledge base and chain-of-thought prompting. Use lambda functions and API calls to enrich results and define instructions and actions.

Test and refine the agent by editing in the agent builder, selecting the Amazon model, and adding interesting facts from the knowledge base, then prepare and add an action group.

Set up action groups to define the agent's tasks, use a lambda function or API schema to fetch exchange rates, define base and target currencies, deploy, test, and review traces.

Create and customize guardrails in Bedrock to enforce safeguards for chatbots and agents, defining blocked prompts, content filters, topics, PII handling, and grounding checks, then test and reuse them.

Test guardrails by selecting models, adding grounding context, and running prompts to verify blocking and masking behaviors, view traces, and assess grounding and relevance scores.

Navigate to your resources and delete agents, knowledge base, and OpenSearch serverless collections to prevent unnecessary costs. Remove related lambda functions and guardrails to complete the cleanup.

Explore the essentials of prompt engineering, defining prompts and the four elements—instructions, context, input data, and output indicator. Learn how negative prompting steers outputs with practical examples such as summarization.

Explore how inference parameters shape model responses by adjusting randomness, diversity, and length, using temperature, top p, and top k settings across Titan and Midas Llama on AWS Bedrock.

Learn how to craft clear prompts with context and directives, ask questions, specify output formats, use prompt templates, and employ trial-and-error to improve model responses.

Explore prompt engineering techniques, including zero shot prompting, few shot prompting, and chain of thought prompting, with detailed explanations and practical examples to guide model outputs.

Explore prompt misuses and risks in ai, including poisoning, hijacking and prompt injection, exposure and prompt leaking, and jailbreaking, and learn how these affect model accuracy, fairness, and reliability.

Explore Amazon Comprehend, a natural language processing service that uses machine learning to extract insights from text, with features like entity recognition, sentiment analysis, and keyphrase extraction.

Explore Amazon Comprehend's capabilities for unstructured text: entity recognition, relationships, sentiment, key phrases, PII, and document classification, including building custom models with labeled training data and API integration.

Explore Amazon Comprehend Medical to extract entities and relationships from unstructured medical text, including medications and dosages, then perform real-time analysis and schedule analysis jobs in S3 buckets.

Explore Amazon Textract, an OCR service that extracts text, handwriting, tables, and forms from documents, with document analysis, custom queries, and APIs for expenses, ID, and lending workflows.

Explore Amazon Textract to extract text and data from documents, including handwriting and scans. Learn to use analyze document API for tables, forms, and signatures, and run custom queries.