If your teenager is between 12 and 18 and has never written a line of AI code, the path from "zero" to "I built something that works" is shorter than most parents think — and faster than waiting until university. Here is the roadmap we use at Novin Zehn for Iranian-diaspora teens worldwide.
Phase 1 — Foundations (1 to 3 months)
Before any AI, your teen needs basic Python: variables, loops, lists, functions, and reading from files. This is also where math anxiety dissolves; Python is the friendliest language for someone who thinks they "aren't a math person." Goal: a small command-line project that takes input and produces output — maybe a quiz game in Persian and English.
Phase 2 — First AI experiments (months 3 to 5)
Now we introduce machine learning using visual and Python-based tools. Teachable Machine for image classification. Scratch-AI for text generation. Then a first real ML model in Python — classifying handwritten digits or predicting house prices from a sample dataset. Concepts: training data, accuracy, overfitting, bias.
Phase 3 — Real projects (months 5 to 9)
This is where their portfolio starts. A computer-vision project that recognizes objects on a webcam. A chatbot trained on their own data. A small recommendation system. They publish their code to GitHub. They write a README. They learn to explain what they did to a non-technical adult — exactly the skill university admissions officers are looking for.
"A 15-year-old with a GitHub portfolio of three working AI projects walks into a Waterloo or TUM interview with more credibility than half the freshmen there." — Novin Zehn senior coach
Phase 4 — Specialization (year 2)
Choosing a focus: computer vision, natural language processing, LLMs, or reinforcement learning. Building a capstone project they can show at a university interview, an internship application, or a tech conference. At Novin Zehn, this is the year our students start preparing for international olympiads and university-level research.
The university angle for diaspora teens
Top CS programs — Waterloo, UofT, McGill, TUM, ETH, MIT — are increasingly portfolio-driven for admissions. A teenager with a GitHub of three real AI projects, a competition certificate, and a clear self-built body of work stands out far more than one with only top grades. For Iranian-diaspora applicants, where admissions committees look hard at evidence of self-driven work, this matters even more.
What gets in the way
- YouTube alone: good for inspiration, fatal as a sole teacher. No feedback loop, no accountability, no community. 90% of teens who try learn-from-YouTube quit within three months.
- Trying to learn the math first: linear algebra before any code? Demotivating. Build first, learn the math when the project demands it.
- No peers: coding alone is much harder than coding alongside others. The Iranian-diaspora teens in our cohorts learn from each other across Toronto, Berlin, and Tehran.
The Novin Zehn teen AI track
Our AI & ML league is built for ages 14–18 and runs in three age tracks. Bilingual instructors, live online classes adapted to your time zone, peer cohorts of Iranian teens in five continents, and a parent panel where you can watch their progress without hovering. The end-of-year showcase is the moment your teen presents to family, peers, and invited guests — the same skill they'll need for their first university interview.
Help your teen build something real
Book a free trial class. We'll match your teen with the right age track and have them building within their first month.
Book your free trial