LLM Challenges in Cyber
Date: Wednesday, January 29, 2025
Time: 17:00-18:30
In an era where Large Language Models (LLMs) are transforming technology, their application in cybersecurity presents unique challenges and opportunities. This webinar features two expert-led sessions that delve into the real-world complexities of leveraging LLMs in high-stakes environments. From designing robust architectures for Retrieval-Augmented Generation (RAG) systems to implementing natural language queries in cybersecurity databases, you’ll gain hands-on insights and practical solutions to address these critical challenges.
*The Webinar will be held in Hebrew
Agenda
17:00-17:45
Building Robust RAG Applications: Architecture Decisions That Matter //Chaim Turkel, Group Leader & Data Architect at Tikal
LLMs are no longer new, but what lessons have we learned over the past year? While building simple applications is relatively straightforward, creating robust and scalable solutions demands thoughtful architectural design.
In this session, Chaim will introduce Ask Tikal, a RAG (Retrieval-Augmented Generation) system developed to query Tikal's internal knowledge base. Leveraging advanced agent techniques, this system evaluates and optimizes results effectively. He will discuss the architectural patterns and agent capabilities that address critical challenges like data leakage, hallucinations, and unreliable outputs—issues that become even more pressing in cybersecurity contexts. By applying these insights, you can unlock the full potential of LLMs for complex, high-stakes applications.
About Chaim
Chaim is a Group Leader at Tikal, with over 20 years of experience in distributed applications and data solutions. He excels in mentoring and guiding teams to achieve their technical and business goals while tackling novel challenges with enthusiasm.
17:45-18:30
Free Text Search Using LLM
Gabi Burabia // Data Science Team Leader at Armis
How do you enable seamless interaction with databases using natural language? Gabi will share how Armis developed a system that translates free-text queries into AQL commands using LLMs. This approach allows users to bypass complex query syntax, opening the door to more intuitive and effective database interactions.
Discover the design choices, challenges, and solutions involved in developing this innovative system, along with its application in cybersecurity to handle sensitive and high-stakes data effectively.
About Gabi
Gabi is a Data Science Team Leader at Armis, managing a multidisciplinary team that drives impactful solutions using LLMs, classification, and anomaly detection. His team’s work focuses on integrating cutting-edge data science techniques into Armis products.
Learn more about Tikal → Here
Learn more about Armis → Here
See you soon!