Under the Hood of an LLM: A Guided Tour from a (Pure Storage) Professor

This month, Lead Principal Technologist and Coffee Break host Andrew Miller welcomes AI practitioner Dr. Robert Alvarez––a teacher and former AI solutions architect for NASA. For those who joined us last November, Robert and Richard are indeed friends, although Robert wholly embraces pineapple on pizza (a controversial subject we’ll be sure to explore).

More seriously, we realized that while there’s a lot of AI news, it’s difficult to see how a large language model (LLM) really works. Yet, there’s real value in that core knowledge to help build your understanding of the overall AI space. Given that, we’ll discuss:

• How models work in a non-mathematical way. Why are we able to use LLMs, and why are they so good?
• Moving beyond language to video from LLMs to LVMs
• Specular decoding, multi-modal challenges, agentic AI, and a bit of DeepSeek commentary