Welcome! 😃 💻🤖

We’re looking forward to a fun few hours with you.  Below are some resources about the upcoming talks and workshop.

<aside> ⭐ Please join the DataStax Discord server prior to the event. That’s where we will be posting important information and updates. Plus, you’ll get to introduce yourself and meet your fellow AI hackers!

</aside>

Demos and Talks

• Gianni Mazza: Overcoming LLM Limitations in Maths and Reasoning for Real-Time Voice Assistants Our AI Maths Tutor utilises structured data to implement Chain of Thought (CoT) reasoning, addressing limitations in LLMs without depending on a specific model. The project leverages a Pub/Sub architecture to decouple components and minimize latency, ensuring responsive voice interactions. By using LangChain to abstract the LLM, we can support multiple models, each with its tailored class for optimal agent configuration and system prompts. We're also experimenting with GPT-o1-preview to generate synthetic data, fine-tuning our own small language model to further reduce costs and latency.
• Lu Wilson and Mini: Teaching models to play tldraw Whenever a new technology appears, our first instinct as developers is to offer text as a way of interacting with it. The same is true for AI. At tldraw, we’ve been working on moving AI interaction away from the chat-based interface, towards a canvas environment. We’ll show the latest from our experiments, and some of the challenges we’ve faced.
• Ahmet Melek: Using Unstructured for an end-to-end RAG Data Setup While most companies pursue RAG capabilities that are powered by internal data, it is no easy feat to consolidate different text data from all the different places they get stored in. In this talk, you’ll learn about the tools that Unstructured provides to unlock all kinds of text data for RAG use, and you'll see the end records in Datastax AstraDB as a proof of end-to-end RAG data enablement procedure. We'll talk about different file types, tables, OCR, sources, chunking, embedding, and destinations. Code samples will be provided for the participants to be able to utilize Unstructured API for a faster RAG development experience.

GenAI Workshop

In the workshop we will build out a RAG bot using DataStax Langflow and Azure OpenAI. By the end you’ll have your own bot that you can experiment with

Getting Ready 🔗

1. Sign Up for Astra if you haven’t already; work email addresses are best!
2. Make sure you have a GitHub account
3. Please Star the LangFlow GitHub to keep up with the latest changes
4. Make sure you have joined the DataStax Discord server
5. The chat interface on GitHub: https://github.com/datastax/langflow-api-chat

Instructions for the workshop

Additional Resources

Pre-Baked Data Sets 💾