The Math Behind AI: Neural Networks and Large Language Models

Invited as a Guest Speaker to VinUniversity, I delivered an advanced technical session focusing on the theoretical underpinnings of modern artificial intelligence. My teaching philosophy heavily emphasizes translating "Theory to Practice." To build robust AI applications, engineers must first understand the fundamental mathematics that drive these systems. This lecture dissected the intricate mathematical logic powering neural networks, specifically exploring the architecture of Large Language Models (LLMs), Transformer models, and Attention mechanisms. By deconstructing the "Math behind AI" for a highly academic audience, the session provided attendees with the deep theoretical foundation required to move past simple API calls and toward designing highly optimized, enterprise-ready Agentic AI orchestrations.