Context Engineering for Engineers

Learn how to engineer context for large language models beyond basic prompts and RAG implementations in this 11-minute conference talk. Discover why context window management is crucial for AI system reliability and performance, exploring how strategic content selection directly impacts output quality. Understand the challenges that arise with long context inputs and why AI performance typically degrades as context length increases. Examine Chroma's technical research findings on context optimization and explore the "needle in a haystack" problem that affects information retrieval in large contexts. Master the "Gather and Glean" methodology for systematically collecting and filtering relevant data before feeding it to language models. Explore practical data gathering techniques and learn how to optimize information through careful gleaning processes. Investigate context engineering applications for AI agents and understand common performance bottlenecks in agent-based systems. Discover compaction strategies that can make AI systems faster and more reliable by reducing context size while preserving essential information. Gain insights from the founder of Chroma on building production-ready AI systems that maintain consistency and effectiveness through thoughtful context management.