🕐
Time
10:00 AM – 11:00 AM CT
👥
Class size
6 students max
Founding cohort · Summer 2026
A launchpad for real AI-powered products.
Every kid your child's age uses AI. Very few understand how it works or how to make it do what they want. Over 8 weeks, each student picks a real problem in their own life, builds an AI-powered app to solve it, and ships it to real users — in a cohort of 6, with guest sessions from the engineers and founders actually building AI in the real world. They leave with a working app, and a mental model of the technology defining their generation that most adults still don't have.
$500 for the founding cohort — $140 off the $640 full price. This rate is only for Summer 2026; future cohorts will be at full price.
S
Your instructor
Shlok S.
CS grad, startup engineer, and builder of several LLM-powered products — teaching the exact workflow he uses to ship AI software professionally every day.
Why this matters
Three reasons this course pays off long after the demo is over.
💡
A real passion project, not a tutorial
Every student picks a real problem in their own life and builds an app for it. Not a canned project from a template — something they actually care about, used by someone they actually know. That kind of ownership is what turns "I took a class" into "I built a thing."
🚀
The skill every job is starting to require
"Can you direct AI to build real things?" is the new "Can you use a computer?" In a few years it'll be on every job description. Students who learn this at 12 aren't cramming for the future — they're starting a decade ahead of their peers.
⚡
Ahead of the ChatGPT crowd
Most of your child's classmates will learn AI as consumers: asking it questions, copy-pasting its answers. Your kid learns to build AI-powered products that other people depend on. That gap between consumer and builder is the one that compounds.
What your student will actually learn
Real concepts, not coding-camp folklore. Every idea is paired with hands-on work inside the student's own project.
01
How LLMs actually work
Next-token prediction, demystified through a 20-minute class game. Most adults using ChatGPT don't have this mental model.
02
Prompting as engineering
Few-shot examples, chain of thought, structured output. Real techniques with measurable effects — not folklore.
03
Evals — how to know it works
The single most underrated AI skill: writing test cases, measuring results, changing one thing, measuring again.
04
AI inside their own app
Calling a model from code, plus tool use. Their app becomes a real AI-powered system — not just one that was built with AI.
05
Why AI fails, and how to debug it
Hallucination, bias, prompt injection. Literacy that'll be as basic as "don't click suspicious links" in five years.
06
The anatomy of a real app
Frontend, backend, database, AI calls. The pieces every app is made of — learned by building their own end to end.
Included in the cohort
Meet the people building AI.
Short, informal sessions with working engineers, quantitative traders, and startup founders — the people shipping the work your child is learning to do. No keynotes, no slides. Just conversations and real questions.
Engineers at
Anthropic
Makers of Claude
From the markets
Quant traders
Startups
VC-backed founders
The 8-week curriculum
One hour live each week, plus light homework. Every session teaches a concept and applies it to the student's own app.
1Week
What is an LLM, really?
Play the next-word prediction game as a class. See that ChatGPT is just that, at scale. Hallucination stops being magic and starts being a consequence of one simple idea.
→ Pick a real, small problem in your life to solve.
2Week
Prompting as specification
Watch a vague prompt and a specific one side by side. Specificity is the whole game. Prompts are instructions, not incantations.
→ Write a spec: what your app does, who uses it, what "working" means.
3Week
First prototype + anatomy of an app
Generate v1 from the spec using AI coding tools. Learn what frontend, backend, database, and AI calls each actually do.
→ Get a working prototype running on your laptop.
4Week
Evals — how do you know it works?
Running something once is not knowing. Measure, change one thing, measure again. The scientific method, applied to AI.
→ Write 5 test cases for your app. Run them. Count pass and fail.
5Week
Putting AI inside your app
Call a language model from code. Your app is now AI-powered, not just AI-built. Intro to tool use — giving the model the ability to do things.
→ Add one AI feature: a summarizer, a chatbot, a classifier.
6Week
When AI fails
Hallucination, bias, prompt injection — demonstrated live. A debugging framework: is it the prompt, the data, your expectations, or the model?
→ Intentionally try to break your own app. Document the failures.
7Week
Memory, context, and shipping
Why the AI forgets. Intuition for RAG without the jargon. How to deploy something to a real public URL for real people.
→ Add memory. Ship the app. Get 3 real users to try it.
8Week
Demo day
Present technical work to a non-technical audience. Parents invited. Every student walks through the problem, the process, the failures, and the working app.
→ 5-minute presentation to the cohort and parents.
What students actually build
Small, personal, and real. The best projects are ones where a specific person in the student's life will actually use the final app.
Grades 5–7
Younger cohort examples
- A flashcard app that turns class notes into practice questions
- A dinosaur (or Pokémon, or Minecraft) facts bot for a younger sibling
- A "what should I wear today" app that checks the weather
- A story generator where they pick the characters and setting
- A chore tracker that writes little encouragements
Grades 7–8
Older cohort examples
- A study helper that generates practice questions from textbook photos
- An AI dungeon master for their D&D group or Discord server
- A Spanish or French conversation partner for test prep
- A journal that reflects on patterns over time
- A debate partner that argues the opposite of any essay prompt
Every student picks their own project on day one. The instructor helps scope it to fit the 8 weeks.
Where the project goes after the course
The 8 weeks are the launchpad, not the finish line. Students leave with a working product and real places to take it.
🚀
Keep shipping
They leave with a working codebase and the muscle memory to add features on their own. Most camps leave kids with something they never touch again. This one doesn't.
🏆
Enter competitions
Congressional App Challenge (middle-school category, fall submission), local hackathons, and school science fair — the project is judging-ready because it has a real user and measured results.
👥
Real users in their life
Their sibling uses the flashcard app. Their robotics team uses the tracker. "X people use this every week" is the real version of "I built an app."
📚
Fall follow-upPlanned
A school-year continuation for students who want to keep building — monthly check-ins, new projects, and a spring showcase. Details coming soon.
Available Sessions
App Incubation — Grades 7-8
June 12 – August 7, 2026
Every Friday, 10:00 AM – 11:00 AM CT
Instructor: Shlok S.
6 spots available
App Incubation — Grades 5-7
June 12 – August 7, 2026
Every Friday, 10:00 AM – 11:00 AM CT
Instructor: Shlok S.
6 spots available