Classroom pilot · one semester · no cost
ML Academy is a free machine learning course that runs entirely in a browser tab, including the Python. Students do not watch lecture videos. They predict before they see the result, then finish each lesson by writing the algorithm and running it.
I am offering a small number of schools and colleges a free one-semester pilot, opened by a live 90-minute session from me.
Five tracks run from no prior knowledge to large language models: foundations, classical machine learning, deep learning, LLMs, and using AI responsibly. The algorithms are not library calls. CART with Gini gain, random forests with bootstrap sampling, an MLP with backpropagation, a BPE tokenizer and the rest are implemented from scratch, so the numbers on screen come out of code the student can read.
Python executes inside the browser tab. There is nothing for a district IT department to install, approve, license or purchase, and it works on a Chromebook.
for step in range(2000):
gw, gb = gradient(w, b)
w = w [ ? ] lr [ ? ] gw
b = b [ ? ] lr [ ? ] gbEvery lesson also asks for a prediction before the simulation runs. The student's hit rate becomes a calibration score: one number showing where their intuition is unreliable. That is the part most courses never measure.
No licence, no purchase order, no vendor onboarding. Students open a link.
Everything runs in the browser tab, including the Python. A Chromebook is enough.
Overfitting, data leakage and statistical testing are taught early, by making students trust a bad number first and then watching it break when the data is re-split.
It sits inside an existing Business & Computer Science pathway, AP CS course or college CIS section rather than displacing anything already scheduled.
One teacher, one section. Thirty minutes on a call is enough to scope it.
Live with the students, in person or over video. I taught this same format most recently as a hands-on workshop at Washington State University.
Self-paced. The first three lessons of every track are open to anyone; a free account, with no card, unlocks the rest and keeps progress across devices.
I ask the teacher what broke and what worked, and I fix it. That feedback is the reason I want a pilot at all.
Including, especially, the parts that did not work.
The name of the school and the number of students, nothing about individuals.
Only if you are willing and only if the pilot went well.
That is the whole ask. There is no paid tier, no upsell and nothing to sign. The course is free permanently, not free for a trial: there is no advertising, no sponsor, no investor and no payment integration of any kind.
I wrote every lesson, every simulation and every verification test in this course, and I give it away. My research is on making AI systems auditable: whether a claim a system makes about its own reasoning can be checked by someone else. It has appeared at the IEEE Conference on Artificial Intelligence (CAI 2026) and IEEE SoutheastCon 2026.
Write to contact@mltraining.org with the name of your school or college, the course it would sit inside, and roughly how many students. I answer every message myself.
You can also just open the course and work through the first lesson before you decide. That takes about ten minutes and tells you more than this page does.