Amazon Bedrock - The Complete Guide to AWS Generative AI

Explore Amazon Bedrock, its supported models, and the Bedrock console view, then set up SDK access and write Python or TypeScript code to access Bedrock models.

Explore how Amazon Bedrock provides an interface to AWS-hosted AI models with serverless on-demand access and various providers. Learn about fine-tuning, model types, and applications like text-to-text and text-to-image generation.

Explore the Amazon Bedrock view in the AWS console, view the model catalog by region, test chat and image models in the playground with on-demand inference.

Create an IAM user for CLI access, install and configure the aws cli, store credentials locally, and use the cli to access bedrock services and list foundation models.

Install and configure the boto3 Python library, set up a virtual environment, and run a Bedrock client to list and inspect foundation models in the proper region.

Install and configure the AWS SDK for JavaScript and TypeScript, set up an npm project, and run Bedrock API calls using the Bedrock client to list and get foundation models.

Learn to configure and use the VS Code debugger for Python and TypeScript, including launch.json setup, virtual environments, breakpoints, and inspecting responses from Bedrock API calls.

Explore bedrock text models, request access, and compare model types and parameters, then practice using the bedrock SDK for text generation in Python and TypeScript.

Learn about AWS Bedrock text models, including output modalities, Nova models, and pricing, and compare invoke API with the unified converse API for easy model switching.

Understand how tokens define input and output limits across text models, how max tokens vary, and how tokenization and costs work using OpenAI's tokenizer.

Explore text model parameters in Amazon Bedrock, including model IDs, inference parameters, top p, temperature, and stop sequences, noting camel case vs snake case across Titan and llama models.

Learn to invoke aws bedrock text models with the python boto3 sdk, using model IDs, region settings, and model-specific request formats for titan text g1 express and llama models.

Learn how to call Bedrock text models with the JavaScript SDK in TypeScript, identify model IDs, and format model-specific requests for Titan and Llama models.

Learn prompt engineering by optimizing inputs to large language models to elicit desired responses. Explore use cases from classifications to summarization and understand prompt anatomy with context, text, and task.

Build a Python chat with history by managing a history array, formatting and injecting history as input, annotating user input with a 'user' prefix, and updating history with model responses.

Add a chat history to a TypeScript chatbot by defining user and bot message types, maintaining a history array, and formatting history for input to preserve context.

Create and configure Bedrock guardrails, define denied topics, set content filters and sample phrases, and test guardrail interventions to protect generative AI prompts.

Attach a guardrail to Bedrock SDK calls by supplying guardrail.config with the guardrail id and a draft version, ensuring regional client instantiation.

Explore image generation with Bedrock models from Amazon Bedrock and learn an introduction to image models. Generate images using the Python or TypeScript SDK, and tune generation with practical parameters.

Explore bedrock image models in the image catalog, with eight models from Amazon and Stability AI; learn how to request access and compare Titan pricing.

Explore Stability AI image models in Amazon Bedrock, focusing on the Stability Image Core v1.1 model and its SDK usage with prompts, inference parameters, and responses, including style preset workaround.

Generate images with stability ai models in the bedrock sdk using python or typescript by providing a prompt to the stable image core v11 model and decoding the base64 image.

Discover how the AWS titan image generator converts text prompts into images, including inpainting, outpainting, and image variation with masks, and configure height, width, and cfg scale.

Dive into editing images with the AWS titan image model. Learn inpainting with mask prompts or mask images to edit regions, test with a cat image, and understand prompt guidelines.

Explore how embeddings power AI understanding inside Amazon Bedrock and use Bedrock models to generate and compare embeddings for words and images using similarity, then learn about vector databases.

Explore embeddings as numerical vectors that group similar data, enabling text and image similarity search, using dot product and cosine similarity with Amazon Titan models.

Learn to generate text embeddings with AWS Bedrock Titan embeddings G1, using Python or TypeScript; input text yields fixed 1536-dim embeddings as long numeric arrays.

Generate embeddings for a list of facts and a new fact, then rank cosine similarities to identify the most relevant fact using Amazon Titan embed text v1 model.

Learn to compute embeddings using the AWS Titan embed text v1 model in TypeScript, compare them via cosine similarity, and identify the most related fact to a new fact.

Compare images with an embedding image model, using the Titan embed image model and cosine similarity on base64-encoded inputs to identify the most similar cat among three samples.

Build an image compare app using image embeddings and cosine similarity to identify the most similar image to a fluffy cat, via base64 encoding and Amazon Titan embed image v1.

Explore vector databases for efficient processing of large embeddings, storing vectors, and querying by similarity using embedding parameters and approximate nearest neighbor search, with examples like pinecone and chroma.

Prepare for hands-on learning with a project-based approach using bedrock models. Build applications and manage infrastructure as code with CDK or Terraform.

Explore building rag applications and integrating Amazon Bedrock with Lang Chain to extract relevant data from local data and a PDF, and chat about pizza, pasta, and weather.

Explore long chain, a framework for building applications powered by large language models and the NFU, with tools, agents, and chains that simplify development and integration with external data sources.

Install dependencies and set up a virtual environment to build a bedrock model with link chain in python, and invoke it with a prompt template to describe a bicycle.

Build a bedrock based generative AI app in TypeScript using lang chain. Install lang chain packages, wrap the bedrock client, and implement a first chain with prompts and model invocation.

Explore retrieval augmented generation concepts and build a RAG app with private or real-time data using embeddings, a vector database, and Lang Chain with Amazon Bedrock, without retraining models.

Build a basic Rag app in python using bedrock, embeddings, lang chain, and an in memory vector store to retrieve the top matching facts and answer user questions.

Build a basic retrieval-augmented generation app in TypeScript using bedrock, LangChain, and an in-memory vector store. Create embeddings, a retriever, and a chat template to answer simple questions.

Load a pdf with py pdf loader into bedrock, embed content, split into chunks with a recursive character text splitter, and query via a LangChain retriever to answer.

Build a pdf rag chat by loading a pdf into a vector store, splitting into paragraphs, retrieving the two most relevant passages, and answering questions with a TypeScript implementation.

Learn to build a summary API using Amazon Bedrock, API Gateway, Lambda, and IAM to summarize long text with a text-to-text model, using Python or TypeScript.

Architect a flow using API gateway, Lambda, and Bedrock to generate a summary from long text with a text-to-text model, and ensure region proximity and IAM role access.

Build a summary lambda in Python using Boto3 and Bedrock, using Titan Text Express v1 in eu central 1. Configure input text and points, invoke the model, and test locally.

Deploy and test an AWS Lambda in the console using Python 3.12 to invoke Bedrock models. Configure permissions, adjust timeouts, and simulate API Gateway requests for Bedrock.

Develop a TypeScript lambda that uses AWS Bedrock Titan Text Express to summarize input text into points via an API gateway, showcasing region-specific configuration and prompt design.

Transpile TypeScript to JavaScript, create and configure a Node.js 20.x AWS Lambda, adjust timeout, grant Bedrock invoke permissions, and test with a proxy event to verify the Lambda response.

Create a rest api in api gateway and connect it to a lambda via proxy integration to summarize text from a post request.

Learn how to implement infrastructure as code with Python and CDK, build a summary API using Lambda, Bedrock, and API gateway, and deploy with CDK synth and CDK deploy.

Build an infrastructure as code text summary API using CDK and TypeScript, including a Lambda, API Gateway integration, Bedrock permissions, and deployment via CDK.

Build an image API using Bedrock API gateway, Lambda, and S3, returning a signed URL from a text description via a POST request in Python or TypeScript.

Outline a serverless flow where an API gateway passes a description to Lambda, which uses Bedrock to generate an image, stores it in S3, and returns a signed URL.

Create a Python image lambda for AWS bedrock, configure an S3 bucket, invoke Titan image generator, save the image to S3, and return a signed URL with local testing.

Create and configure a Python 3.12 lambda in the AWS console, set the handler and timeout, grant Bedrock and S3 permissions, and test end-to-end image generation.

Build a lambda that invokes bedrock's Titan image generator model to create an image, save it to S3, and return a signed URL.

Expose AWS bedrock image generation through API gateway with lambda proxy integration, using a post request with a description body, deploy to prod, and test via REST client tools.

Learn infrastructure as code with AWS CDK in Python to build an image API using an S3 bucket, Lambda, API Gateway, and Bedrock invocation, deployed with CDK synth and deploy.

Build an image api using infrastructure as code with aws cdk and typescript, creating an s3 bucket, a node.js lambda, and an api gateway.