Microsoft Reactor

Join us for the Reactor Series - GenAI for Software Developers V2, where we delve into the cutting-edge world of Large Language Models (LLMs) and their applications. Across four illuminating episodes, we'll explore the intricacies of LLM evaluation, their groundbreaking potential in healthcare, and revolutionary natural language interfaces for analytics.

About the Series:
Through a series of engaging sessions, participants will gain insights into cutting-edge technologies, best practices, and real-world use cases, empowering them to leverage AI to drive innovation in their own projects.

Who Should Attend:
This session is ideal for software developers, data scientists, AI enthusiasts, and anyone interested in the intersection of artificial intelligence and software development. Whether you're looking to stay updated on the latest trends in AI or seeking practical insights to enhance your development projects, this session offers valuable knowledge and perspectives.

Why Should You Join:

• Gain insights into the evaluation methods of Large Language Models (LLMs) from an industry expert.
• Explore the significance of LLM assessment and its implications for responsible development.
• Learn about core capabilities, challenges, and methods used to ensure the effectiveness of LLMs.
• Network with fellow software developers and AI enthusiasts to exchange ideas and best practices.
• Stay updated on the latest advancements in artificial intelligence and their applications in software development.

Agenda:

Session Title: Kusto Copilot
Speaker: Adi Eldar Principal Data Scientist

Session Overview:
In this talk we are going to present Kusto Copilot in Microsoft Fabric. Through a sidebar interface, Fabric users could write their analytics requirements in natural language and the Copilot will generate the required KQL, to be confirmed by the user and run. We are going to deep dive into the the R&D that was done for high quality NL2KQL generation. First we will walk through the current NL2KQL architecture that maximize the capabilities of OpenAI models by advanced prompt engineering (RAG based), and post-process to validate/fix the generated KQL. Then we will explore the research framework that we built for acquiring labeled data sets, running A/B tests to compare different LLM flows and how we measure and compare the accuracy of different flows in order to chose the best one based on large statistics.

Join us as we unravel the intricacies of LLMs and their transformative potential in software development. Don't miss out on this enlightening series at Reactor Series - GenAI for Software Developers V2.