Talk Details
Title
Unlock Generative AI for Swift Server-Side Development
Description
In the rapidly evolving landscape of artificial intelligence, Swift developers are looking to integrate powerful language models into their applications. While AWS's Amazon Bedrock provides access to leading LLMs and generative AI capabilities, working with these services in strongly-typed languages like Swift presents unique challenges. The varying payload structures across different LLMs, combined with automatically generated SDKs that don't always align with Swift's elegant syntax patterns, can make implementation complex. This created an opportunity to develop a more Swift-idiomatic approach to working with Bedrock. So, I created the Swift Bedrock Library, that forms a layer on top of the AWS SDK for Swift for interacting with Amazon Bedrock. This library provides a convenient way to access Amazon Bedrock's capabilities from Swift applications. By providing clear, concise, and Swift-specific implementations, it accelerates the learning curve and enables developers to quickly take advantage of Bedrock's advanced models. From streaming chat to image generation or variation, the library covers a wide range of use cases. I also worked on the Swift Foundation Model Playground, an example web app that demonstrates the use of the Swift Bedrock Library. It is built with a front-end in React and a backend that uses Hummingbird. To help out developers that prefer to use the AWS SDK for Swift as is, I created the very first Swift code examples for Bedrock Runtime. These are now a part of the official AWS documentation and include chatting with Amazon Nova, Meta Llama and Anthropic Claude both with and without streaming responses. During my talk I will share the process I went through to during my internship and how the work I did can improve your experience when you decide to implement AI in your next project.
About the Speaker
Position
Applied Informatics graduate
Bio
Hi there! I am Mona, a recent Applied Informatics graduate who is passionate about the possible ways to apply AI to solve real-life problems.
