SouthTech AI & Data meetup

Join us at the SouthTech AI & Data Collective - a boutique gathering of data enthusiasts and AI professionals in Southern Israel. Discover cutting-edge trends in data science, machine learning, and artificial intelligence in an intimate, collaborative setting that encourages meaningful connections and innovative conversations. Whether you're a seasoned expert or simply passionate about the field, this event is crafted to inspire, educate, and ignite collaboration across the AI and data community.

17:00 - 17:30 Gathering & Networking

17:30 -18:15 The Power and Pitfalls of Data in AI: How LLMs Learn, Thrive, and Fail by Yair Nevet (Principal Engineer at LSports)

18:15 - 18:45 Unlocking the Power of Data+AI : Building Data Intelligent AI Systems by Michael Krasnopolsky (Solutions Architect at Databricks)

18:45 - 19:00 Break & Networking

19:00 - 19:30 PoC to Production: Building Scalable AI Applications by Nir Shtein (CTO of Sherpa.io)

Sessions details:

The Power and Pitfalls of Data in AI: How LLMs Learn, Thrive, and Fail by Yair Nevet (Principal Engineer at LSports)

Large Language Models (LLMs) have revolutionized the way we interact with AI, but their outputs are far from neutral or static. In this talk, we’ll dive into a fundamental truth: LLMs are shaped by the data they are trained on, which means their responses can sometimes reflect biases, stereotypes, and even societal trends.

We’ll explore how these biases emerge, why LLMs are probabilistic rather than deterministic, and how their behavior evolves over time as data shifts. We'll also discuss real-world implications—how these models can reinforce existing narratives, where they might fall short, and what we can do to mitigate biases while working with them.

Unlocking the Power of Data+AI : Building Data Intelligent AI Systems by Michael Krasnopolsky (Solutions Architect at Databricks)

Everybody are amazed by the leap we made with GenAI, but ~70-85% of GenAI project fail to reach production.
We'll delve into the key concepts and strategies for harnessing the full potential of your data and AI to solve enterprise business challenges.
First, we'll examine the distinction between general intelligence and data intelligence, and why the latter is crucial for organizations seeking to solve business problems. We'll then introduce the concept of a Data Intelligence Platform, and discuss its role in enabling the development of AI-driven applications.
Next, we'll outline the essential pillars to consider when building an AI Agentic System, providing a comprehensive framework for designing and implementing intelligent systems that can drive real business value.
To bring these concepts to life, we'll also showcase a live demo, taking you on a journey from prototyping to deployment, and highlighting the key considerations and best practices for getting your AI projects up and running quickly and effectively.
By the end of this presentation, you'll have a deeper understanding of what it takes to build Data Intelligent AI Systems, and be equipped with the knowledge and insights needed to drive your own AI initiatives forward.

PoC to Production: Building Scalable AI Applications by Nir Shtein (CTO of Sherpa.io)

Turning a Proof of Concept (PoC) into a production-ready LLM application is no small feat. This talk dives into the challenges of building scalable LLM solutions, focusing on critical considerations such as modular design, maintainability, and handling unexpected model behaviors like hallucinations. We’ll explore the nuances of integrating LLMs into existing systems, balancing flexibility with stability, and avoiding common pitfalls that can lead to unreliable outputs. Additionally, we’ll discuss effective strategies for testing and validating LLM responses to ensure accuracy and consistency. Whether you're starting out or looking to optimize an existing application, this session will equip you with practical insights and best practices for delivering robust LLM solutions in production.