Numeria
I am building a maths game for my daughter: short lessons, useful practice, and a world that gives her a reason to come back. The path runs from missing foundations toward calculus.
Why I built itJonathan Laliberte-Alle
Maths, explained around the kitchen table.
I'm Jonathan: a quant, teacher, and dad. At home and at Primoris Academy, I look for the explanation that makes a particular child say, “Oh. That makes sense.” Sometimes it is a picture. Sometimes it is a spreadsheet. Sometimes it is an entire game I built because I wanted to impress my daughter. Dad Maths is where I keep the good bits and share them with parents and teachers.
I am still building this place. Lesson notes, classroom materials, and the occasional useful failure will appear as they are ready.
What is here
A couple of projects, plus notes from the classroom.
Most of this starts with one child or one stubborn question. I try an explanation, build something when words are not enough, and share the useful bits with parents, teachers, and fellow builders.
Teaching
Core maths at each student's pace, plus finance and data science taught through problems worth arguing about.
Explore the coursesBuilding with AI
What happens when a young person who can build with AI sits down with an adult who knows a real problem inside out?
Read the premise
Early work
AI-assisted reverse mentoring
Put a young AI builder next to an adult with a real problem.
My starting point is simple: children should learn to reason with AI, not let AI reason for them. I am beginning to teach 10- to 14-year-olds how to turn a real problem into a useful prototype while checking the model's work and keeping human judgment in charge.
The experiment
An adult brings experience and a problem from real life. A young builder asks questions, maps the work, and uses AI to make a first version. Together they test whether it is accurate, useful, safe, and understandable.
This reverses the usual mentoring relationship without pretending that age or technical fluency replaces expertise. The adult knows the domain. The student learns how to build. Each has something the other needs.
The work should be supervised by a parent or teacher and use fictional or sanitized examples. Children should never put private family, school, client, or business information into an AI system.
What students should learn
- Find a problem that is specific enough to solve and important enough to matter.
- Interview the person who understands the work instead of guessing what they need.
- Break a solution into steps, give AI clear constraints, and revise the result.
- Check facts, calculations, edge cases, privacy, and failure modes before trusting a prototype.
- Explain what they built, where it is reliable, and where a person must remain responsible.
A student who can interview an expert, draft a prototype, and test it carefully can contribute to real work now. The educational job is to pair that new ability with judgment.
About me
I have been doing some version of this since I was seven.
Mathematics and computer science have been part of my life since I was seven. I studied mathematical engineering at Télécom Paris and probability and finance at Université Pierre et Marie Curie and École Polytechnique, then spent my career building systematic trading and execution systems.
Becoming a parent made the question practical: how do you help a child get unstuck without stealing the satisfying part? Teaching at Primoris Academy, building Numeria, and training Brazilian jiu-jitsu with my family are now different versions of that question.