Presentation

​Context Is the New King: Overcoming Contextual Challenges in LLM Applications

​Abstract
As Large Language Models (LLMs) evolve from simple chatbots to powering complex, production-grade systems, one challenge consistently rises to the top: managing context. Whether it's a single-turn query or a multi-step agentic workflow, the ability to maintain and manipulate context is critical to building reliable, intelligent applications.

​This talk introduces the emerging discipline of context engineering—a set of principles and practices for designing LLM systems that are robust, scalable, and context-aware. We will explore real-world challenges and practical strategies for handling context in LLM-based applications, including agent-based architectures where LLMs must reason, plan, and act over time.