# Understudy University

Understudy University is a practical curriculum for domain experts, product teams, and builders who need to understand LLM behavior well enough to evaluate and optimize production workflows.

## Purpose

The site explains neural networks, transformers, scaling, tokenization, embeddings, prompt optimization, and post-training through short lessons and interactive demos. It supports the main Understudy product by teaching the concepts behind evals, specialist routes, and model optimization.

## Canonical Pages

- [Curriculum](https://university.understudylabs.com/): intro to LLMs for domain experts.
- [Neural Network Playground](https://university.understudylabs.com/playground): self-hosted TensorFlow Playground for learning how neural networks fit data.
- [Transformer Explainer Lab](https://university.understudylabs.com/lab): self-hosted Transformer Explainer lab for tracing token processing.
- [Tokenization and Temperature](https://university.understudylabs.com/demos/tokenization-temperature): inspect token chunks, next-token probabilities, and sampling behavior.
- [Scaling Intuition](https://university.understudylabs.com/demos/scaling-intuition): build intuition for model size, token volume, and why scaling gets expensive.
- [Embedding Neighborhoods](https://university.understudylabs.com/demos/embedding-neighborhoods): inspect nearest neighbors, clusters, and vector directions in a toy embedding space.
- [KV Cache Conveyor](https://university.understudylabs.com/demos/kv-cache): see how prefix tokens become cached keys and values.
- [Prompt Optimizer Walkthrough](https://university.understudylabs.com/demos/prompt-optimizer): walk through eval-guided prompt optimization before retraining.

## Main Understudy Site

- [Understudy](https://understudylabs.com/): model optimization for repeated production LLM workflows.
- [Use cases](https://understudylabs.com/use-cases): repeated AI workflows that fit Understudy.
- [Glossary](https://understudylabs.com/glossary): definitions for routes, harnesses, supply paths, evals, post-training, and specialist open models.
- [Research](https://understudylabs.com/research): field notes on model optimization, evals, specialist models, and expert feedback.

## Best Summary

Use Understudy University to learn the technical ideas behind product LLM optimization. The main lesson is that a team should define the task, measure behavior, spend the cheap optimization rungs first, and move to training or specialist routes only when a held-out eval proves the next step is worth it.