Night School
Under the Hood of an LLM: How Neural Networks Actually Work (Part 1 of 2)
About this lecture
Most engineers got into AI coding without ever learning what the model is doing. You can ship with a coding agent all day and still have no mental picture of what happens between your prompt going in and the tokens coming out. This session is the crash course. Aaron Gallant walks through how text becomes vectors and embeddings and gets translated back out, how a neural network is actually structured, and how the foundations of classical machine learning carry directly into the LLMs you use every day. He will also make the surrounding vocabulary readable — training pipelines, hyperparameters, fine-tuning, reinforcement learning — so the field stops feeling like it belongs only to researchers with PhDs. This is part one of a two-part September series. You will leave with a repo of notebooks to work through before part two, where we build a small neural network from scratch.
This Night School runs in two parts — same teacher, two sessions two weeks apart, with work to do in between. Part one on September 2 covers what is actually happening inside the model; part two on September 16 builds one from scratch. You will get a repo of notebooks after the first session, so you come back having built something rather than having watched something. Register for both.
Note on prerequisites: this is a crash course, not an introduction to programming. If you are already a little bit technical you will be fine. If you have never written code, this one will be a hard start.
In this lecture
- Understand what is actually happening inside a model when your text is turned into vectors and embeddings and translated back out
- Get working definitions for the terms that gate the field: training pipelines, hyperparameters, fine-tuning, reinforcement learning, and RL environments
- See how the foundations of classical machine learning translate into today’s large language models
- Learn why the barrier to training a model is mostly framework and vocabulary, not budget or credentials
- Leave with a repo of notebooks to work through before part two
About the instructor
Aaron Gallant — Lead Instructor at Gauntlet AI, with Gauntlet since the beginning. He has spent 18 years in technology across software development, data science and analysis, data pipelines, and internal tooling, including five years at Google building tools for internal support workflows and operations. He works on on-prem models and has run training pipelines himself, including generating synthetic data to train HIPAA-compliant models for clinical research. He holds a master’s degree in Computer Science from the University of Illinois at Urbana-Champaign and a bachelor’s degree in Political Science and Philosophy from the University of Rochester. As Gauntlet AI CTO Ash Tilawat put it: “If there’s one person that truly understands the foundations of classical ML and how they translate into LLMs, that’s Aaron.”