As AI systems become foundational infrastructure, developers must adapt to building with components that are probabilistic, emergent, and often opaque. This talk distills practical lessons from deploying agentic architectures and benchmarking LLM capabilities, focusing on how to test, debug, and reason about non-deterministic behavior. We’ll explore infrastructural patterns, effective evaluation strategies, and the growing role of soft skills—like abstraction and linguistic precision—in prompt engineering. Aimed at bridging the gap between traditional development and AI-native thinking, this session offers a pragmatic guide to engineering with intelligence.
Igor Ševo is a computer scientist, writer, and theorist with a Ph.D. in machine learning. As Head of Artificial Intelligence at HTEC Group, he leads initiatives in agentic systems, LLM integration, and AI infrastructure. His interdisciplinary work spans software engineering, cognitive science, and philosophy, with a focus on understanding and building intelligent systems. Igor has authored several...