AWS Generative AI CDK Constructs

All classes are under active development and subject to non-backward compatible changes or removal in any future version. These are not subject to the Semantic Versioning model. This means that while you may use them, you may need to update your source code when upgrading to a newer version of this package.

---

Table of contents

• Introduction
• CDK Versions
• Contributing
• Design guidelines and Development guide
• Getting Started
• Catalog
• Sample Use Cases
• Additional Resources
• Contributors
• Operational Metrics Collection
• Roadmap
• License
• Legal Disclaimer

Introduction

The AWS Generative AI Constructs Library is an open-source extension of the AWS Cloud Development Kit (AWS CDK) that provides multi-service, well-architected patterns for quickly defining solutions in code to create predictable and repeatable infrastructure, called constructs. The goal of AWS Generative AI CDK Constructs is to help developers build generative AI solutions using pattern-based definitions for their architecture.

The patterns defined in AWS Generative AI CDK Constructs are high level, multi-service abstractions of AWS CDK constructs that have default configurations based on well-architected best practices. The library is organized into logical modules using object-oriented techniques to create each architectural pattern model.

CDK Versions

AWS Generative AI CDK Constructs and the AWS CDK are independent teams and have different release schedules. Each release of AWS Generative AI CDK Constructs is built against a specific version of the AWS CDK. The CHANGELOG.md file lists the CDK version associated with each AWS Generative AI Constructs release. For instance, AWS Generative AI CDK Constructs v0.0.0 was built against AWS CDK v2.96.2. This means that to use AWS Generative AI CDK Constructs v0.0.0, your application must include AWS CDK v2.96.2 or later. You can continue to use the latest AWS CDK versions and upgrade the your AWS Generative AI CDK Constructs version when new releases become available.

Contributing

Contributions of all kinds are welcome! Check out our contributor guide

Design guidelines and Development guide

If you want to add a new construct to the library, check out our design guidelines, then follow the development guide

Getting Started

TypeScript

• Create or use an existing CDK application in TypeScript.
  • cdk init app --language typescript
• Run npm install @cdklabs/generative-ai-cdk-constructs
• The package should be added to your package.json.
• Import the library:
  • import * as genai from '@cdklabs/generative-ai-cdk-constructs';

Python

• Create or use an existing CDK application in Python
  • cdk init app --language python
• Install the package:
  • pip install cdklabs.generative-ai-cdk-constructs
• Import the library:
  • import cdklabs.generative_ai_cdk_constructs

NuGet

• Create or use an existing CDK application in Python
  • cdk init app --language csharp
• Install the package while in the Visual Studio project:
  • dotnet add package CdkLabs.GenerativeAICdkConstructs
• Use the namespace:
  • using Cdklabs.GenerativeAiCdkConstructs;

Go

• Create or use an existing CDK application in Python
  • cdk init app --language go
• Get the module:
  • go get github.com/cdklabs/generative-ai-cdk-constructs-go/generative-ai-cdk-constructs
• Import the library:
  • import "github.com/cdklabs/generative-ai-cdk-constructs-go/generative-ai-cdk-constructs"

NOTE: The Go distribution repository, distributes the JSII tar gzipped versioned source from the source repository

Java

• Create or use an existing CDK application in Java
  • cdk init app --language java
• Add the dependency into the pom.xml

<dependency>
    <groupId>io.github.cdklabs</groupId>
    <artifactId>generative-ai-cdk-constructs</artifactId>
    <version>Get the latest version and insert it here</version>
</dependency>

Refer to the documentation for additional guidance on a particular construct: Catalog

Catalog

The following constructs are available in the library:

L3 constructs

Construct	Description	AWS Services used
Data ingestion pipeline - OpenSearch	Ingestion pipeline providing a RAG (retrieval augmented generation) source for storing documents in a knowledge base.	Amazon OpenSearch, AWS Step Functions, Amazon Bedrock, AWS AppSync, AWS Lambda
Question answering	Utilizing Large Language Models (Anthropic Claude V2.1.) for Question Answering on PDF documents with RAG (retrieval augmented generation) source and/or long context. Additionally, leveraging Anthropic Claude 3 for visual question answering on images.	Amazon OpenSearch, AWS Lambda, Amazon Bedrock, AWS AppSync
Summarization	Document summarization with a large language model (Anthropic Claude V2.1).	AWS Lambda, Amazon Bedrock, AWS AppSync and Amazon ElastiCache for Redis.
SageMaker model deployment (JumpStart)	Deploy a foundation model from Amazon SageMaker JumpStart to an Amazon SageMaker endpoint.	Amazon SageMaker
SageMaker model deployment (Hugging Face)	Deploy a foundation model from Hugging Face to an Amazon SageMaker endpoint.	Amazon SageMaker
SageMaker model deployment (Custom)	Deploy a foundation model from an S3 location to an Amazon SageMaker endpoint.	Amazon SageMaker
Content Generation	Generate images from text using Amazon titan-image-generator-v1 or stability.stable-diffusion-xl-v1 model.	AWS Lambda, Amazon Bedrock, AWS AppSync
Amazon Bedrock Monitoring (Amazon CloudWatch Dashboard)	Amazon CloudWatch dashboard to monitor model usage from Amazon Bedrock.	Amazon CloudWatch
TXT to SQL	Leverages generative AI capabilities to facilitate natural language-based SQL query generation.	Amazon Event Bridge, Amazon Bedrock, AWS Lambda, Amazon SQS, AWS Secrets, and database of choice
LlamaIndex Data Loading	Use LlamaIndex to load data in preparation for generative AI workloads	Amazon ECS Fargate, Amazon SQS, and AWS Systems Manager Parameters

L2 Constructs

Construct	Description	AWS Services used
Lambda layer	Python Lambda layer providing dependencies and utilities to develop generative AI applications on AWS.	AWS Lambda, Amazon Bedrock, Amazon SageMaker
Amazon Bedrock	CDK L2 Constructs for Amazon Bedrock.	Amazon Bedrock, Amazon OpenSearch Serverless, AWS Lambda
Amazon OpenSearch Serverless Vector Collection	CDK L2 Constructs to create a vector collection.	Amazon OpenSearch Vector Index
Amazon OpenSearch Vector Index	CDK L1 Custom Resource to create a vector index.	Amazon OpenSearch Serverless, AWS Lambda

Sample Use Cases

The official samples repository includes a collection of functional use case implementations to demonstrate the usage of AWS Generative AI CDK Constructs. These can be used in the same way as architectural patterns, and can be conceptualized as an additional "higher-level" abstraction of those patterns. Those patterns (constructs) are composed together into stacks, forming a "CDK app".

Additional Resources

Resource	Type	Description
AWS re:Invent 2023 - Keynote with Dr. Werner Vogels	Keynote	Dr. Werner Vogels, Amazon.com's VP and CTO, announces the AWS Generative AI CDK Constructs during his AWS re:Invent 2023 keynote.
Workshop - Building Generative AI Apps on AWS with CDK	Workshop	In this workshop, you will explore how to build a sample generative AI app on AWS using CDK and Generative AI CDK Constructs.
Workshop - Hands on AWS CDK Generative AI Constructs	Workshop	In this workshop you will deploy projects that use CDK constructs from this library. Projects are from the amazon-bedrock-samples Github Repository.
Build generative AI applications with Amazon Titan Text Premier, Amazon Bedrock, and AWS CDK	Blog post + Code sample	Blog post exploring building and deploying two sample applications powered by Amazon Titan Text Premier using the Generative AI CDK constructs.
aws-cdk-stack-builder-tool	Code sample	AWS CDK Builder is a browser-based tool designed to streamline bootstrapping of Infrastructure as Code (IaC) projects using the AWS Cloud Development Kit (CDK).
CDK Live! Building generative AI applications and architectures leveraging AWS CDK Constructs!	Video	CDK Live! episode focused on building and deploying generative AI applications and architectures on AWS using the AWS Cloud Development Kit (CDK) and the AWS Generative AI CDK Constructs.
Announcing AWS Generative AI CDK Constructs!	Blog post	Blog post announcing the release of the AWS Generative AI CDK Constructs.
Streamline insurance underwriting with generative AI using Amazon Bedrock	Blog post + Code sample	Blog post and code sample discussing how to use AWS generative artificial intelligence (AI) solutions like Amazon Bedrock to improve the underwriting process, including rule validation, underwriting guidelines adherence, and decision justification.
aws-genai-llm-chatbot	Code sample	Multi-Model and Multi-RAG Powered Chatbot Using AWS CDK on AWS allowing you to experiment with a variety of Large Language Models and Multimodal Language Models, settings and prompts in your own AWS account.
bedrock-claude-chat	Code sample	AWS-native chatbot using Bedrock + Claude (+Mistral).
amazon-bedrock-rag	Code sample	Fully managed RAG solution using Knowledge Bases for Amazon Bedrock.
Amazon Bedrock Multimodal Search	Code sample	Multimodal product search app built using Amazon Titan Multimodal Embeddings model.
Amazon Bedrock Knowledge Bases with Private Data	Blog post + Code sample	Blog post and associated code sample demonstrating how to integrate Knowledge Bases into Amazon Bedrock to provide foundational models with contextual data from private data sources.
Automating tasks using Amazon Bedrock Agents and AI	Blog post + Code sample	Blog post and associated code sample demonstrating how to deploy an Amazon Bedrock Agent and a Knowledge Base through a hotel and spa use case.
Agents for Amazon Bedrock - Powertools for AWS Lambda (Python)	Code sample	Create Agents for Amazon Bedrock using event handlers and auto generation of OpenAPI schemas.
Text to SQL Bedrock Agent	Code sample	Harnessing the power of natural language processing, the "Text to SQL Bedrock Agent" facilitates the automatic transformation of natural language questions into executable SQL queries.

Contributors

Operational Metrics Collection

Generative AI CDK Constructs may collect anonymous operational metrics, including: the region a construct is deployed, the name and version of the construct deployed, and related information. We may use the metrics to maintain, provide, develop, and improve the constructs and AWS services.

Roadmap

Roadmap is available through the GitHub Project

License

Apache-2.0

Legal Disclaimer

You should consider doing your own independent assessment before using the content in this library for production purposes. This may include (amongst other things) testing, securing, and optimizing the CDK constructs and other content, provided in this library, based on your specific quality control practices and standards.

---

© Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.