AI CLASS / LESSONS 4–9
Language models and building with AI
Explore how language models and coding agents work, then plan and build a small app of your own.
These lessons continue our classifier, reinforcement learning, and clustering labs. The assignments use the VCC coding classroom ↗.
What is an LLM?
An LLM predicts the next token from its input. Attention, neural network layers, and chat formatting explain how it produces a conversation.
Open lesson →LESSON 5 · 50–60 minutesHarnesses
A harness manages the conversation with an LLM and runs its tools. The assignment is to build a harness in VCC and run it in the terminal.
Open lesson →LESSON 6 · 50–60 minutesMixture of Experts
An agent can route a task to a tool or another agent. This lesson compares language predictions with code and explains how that routing differs from a model’s internal Mixture of Experts.
Open lesson →LESSON 7 · 45–50 minutesHow a model builds a response
Follow a sentence through a small neural network, watch its numbers change, and investigate why changing the context can change what comes next.
Open lesson →LESSON 8 · 45–50 minutesHow MoE shares the work
Watch a router choose clusters of neurons. Compare using every cluster with using a small group, then grow the model and see which kinds of work increase.
Open lesson →LESSON 9 · 55–65 minutesVibe coding and project thinking
Choose a problem someone has, describe a small app that could help, and work with a coding agent to build and check its first version.
Open lesson →