Building a RAG Pattern chat bot with Azure OpenAI and LangChain.js | Azure Developers JavaScript Day

 

The Confluence of Expertise

 

Lars introduced his extensive background, highlighting his contributions to Developer Technologies, Open-Source projects, and educational initiatives. His foundational work with the This is Learning community, which evolved from an Angular-centric hub to a broader tech community, showcases his commitment to fostering growth and knowledge sharing among tech enthusiasts of all skill levels.

 

Wassim and Natalia, complemented the discussion with their vast experiences. Their collaboration has led to the development of samples that guide the community on best practices for application implementation using Azure services and Developer Tools.

 

You can check out Wassim's profile on GitHub and Natalia's on GitHub. All the code and projects they develop are available there!

 

 

Wassim and Natalia walked through the architecture of the sample RAG chatbot application, detailing the utilization of various Azure services, including:

 

• Azure Static Web Apps for the UI deployment
• Azure Container Apps for the backend logic
• Azure AI Search for data indexing and retrieval

 

The integration with OpenAI’s APIs plays a pivotal role in fetching predictive models and responses.

 

The session also shed light on the application's backend components: the indexer and the search API, highlighting the use of Fastify for API exposure and the strategic implementation of the sliding window algorithm for efficient text chunking and indexing.

 

Below you can see a diagram of the project's architecture:

 

 

 

The practical segment involved a detailed code walkthrough, emphasizing npm workspaces, monorepo organization, and the use of libraries like LangChain.js for seamless interaction with Azure OpenAI. The discussion also touched on the importance of semantic search and vector embeddings in improving the chatbot's response quality.

 

 

 

You can get an overview of the project's code at GitHub. Help us improve the project with your contributions and feedback! The project is completely open-source and available to the community. And give stars to the repository if you like it! ;)

 

 

 

Lars provided a comprehensive overview of the latest features and improvements in Angular, including:

 

• Signals for managing state changes.
• Standalone components for modular development.
• And the evolution of template syntax for more intuitive code structuring.

 

The session illustrated how these Angular advancements could be leveraged in the context of the RAG chatbot application, enhancing its functionality and user interaction.

 

You can see in the image below an example of how the chatbot has been implemented in the user interface, making use of Angular's new features:

 

 

 

The deployment process to Azure was demystified, showcasing the use of:

 

• Azure Developer CLI: which is a command line tool that allows you to create, manage and deploy Azure resources directly from the terminal.

• Bicep: which is an Infrastructure as Code (IaC) language that allows Azure resources to be defined declaratively.

The open invitation for community contributions to the project underscores the collaborative spirit of the initiative.

 

 

This enriching session at the JavaScript Developer Days 2024 not only offered a comprehensive guide to developing a RAG pattern chatbot but also illuminated the potential of combining Angular’s new features with Azure's robust cloud services. The contributions of Lars, Wassim, and Natalia provided attendees with valuable insights into leveraging modern web development practices and AI technologies to create dynamic, responsive applications.

 

The collaborative effort to build and refine this project invites developers from across the spectrum to contribute, learn, and grow within this innovative domain, further bridging the gap between AI advancements and practical web development applications.

 

Additional Resources

 

• Free Course - Learn Path - Bicep
• Free Course - Implement Retrieval Augmented Generation (RAG) with Azure OpenAI Service 
• Collection - Microsoft Ignite: Generative AI with Azure OpenAI
• Collection - Master the basics of Azure: AI Fundamentals
• Free Course - Using GitHub Copilot with JavaScript
• Free Course - Build a Form Recognizer custom skill for Azure Cognitive Search
• Free Course - Search Azure Cosmos DB for NoSQL data with Azure Cognitive Search
• Free Course - Azure Static Web Apps Learn Path