This library provides AWS Cloud Development Kit (CDK) constructs for defining common generative AI infrastructure patterns in code. It targets developers building generative AI solutions on AWS, enabling the rapid, repeatable deployment of well-architected, multi-service architectures.

How It Works

The constructs are high-level, object-oriented abstractions of AWS CDK constructs. They encapsulate best-practice default configurations for common generative AI patterns, simplifying the creation of complex AWS infrastructure. The library is organized into logical modules, promoting code reusability and maintainability for generative AI workflows.

Quick Start & Requirements

Installation varies by language:

• TypeScript: npm install @cdklabs/generative-ai-cdk-constructs
• Python: pip install cdklabs.generative-ai-cdk-constructs
• C#: dotnet add package CdkLabs.GenerativeAICdkConstructs
• Go: go get github.com/cdklabs/generative-ai-cdk-constructs-go/generative-ai-cdk-constructs
• Java: Add dependency to pom.xml.

Requires a compatible AWS CDK version (check CHANGELOG.md). Official documentation and workshops are available.

Highlighted Details

• Provides constructs for SageMaker model deployment (JumpStart, Hugging Face, Custom).
• Includes patterns for Amazon Bedrock integration, including monitoring and batch inference.
• Offers constructs for vector collection and indexing with Amazon OpenSearch Serverless.
• Sample use cases and workshops demonstrate practical application.

Maintenance & Community

Contributions are welcome. A roadmap is available via GitHub Projects.

Licensing & Compatibility

Licensed under Apache-2.0.

Limitations & Caveats

All classes are under active development and subject to non-backward compatible changes or removal without notice, not adhering to Semantic Versioning. Users may need to update source code when upgrading. An independent assessment for production use is recommended.