Train-your-own tiny LLM
A miniature language model trained on data your teen chooses, with before-and-after results.
Open the black box, one word at a time.
ChatGPT is not magic, it is math your teen can genuinely understand. In this course they train a tiny language model of their own and watch it learn, word by word, inside our simulations.
Light Python is introduced along the way. Curiosity matters more than math level.
Every unit is taught inside a live simulation and ends with something your child has actually built. Pace is 1-on-1, so no one is rushed or held back.
Tokens: how a sentence turns into something a computer can compute with.
Next-word prediction, played by hand first, then handed to the machine.
Feed a tiny model your own data and watch its answers improve in real time.
The trick that lets a model keep track of what matters across a whole sentence.
Hallucination, bias, and why the model sounds certain exactly when it should not.
Kids remember what they make. Each project is a genuine artifact your child keeps, shares and builds on.
A miniature language model trained on data your teen chooses, with before-and-after results.
A short write-up documenting where their model fails and why, like a real researcher.
A clear explainer that debunks a common AI headline using what they now understand.
We keep it intuition-first. Every idea starts as a hands-on game before any math or code appears, so curious 11-year-olds thrive alongside teens.
Comfort with basic arithmetic is plenty. New ideas are introduced visually, exactly when they are needed, never as a wall of equations.
Yes, a genuinely small language model in our lab. It is simple enough to understand end to end, but it truly learns from data they provide.
AI Agents & Automation for the builders, or our Machine Learning course for teens who want the full engineering path.
Meet the mentor, play inside a real simulation, and watch your child build something on day one. No card required.