AI coding for Hong Kong schools, including international schools
Vibe Coding × AI LLM
Write programmes in natural language · Understand large language models
VibeCode HK is for secondary students, upper primary, and schools—including international schools—that want to launch AI classes. We move generative AI from “opening a chat” to a teachable path: foundations first, then prompts, then Vibe Coding into Python projects.
The course centres on three promises—understand · produce · take away: students can explain how an LLM works, write usable prompts and demonstrable projects, and leave with ethics awareness and showcase directions.
Three commitments run through the course: explain foundations, write and verify by hand, and take away a showcase-ready project.
01
Students can explain how an LLM “reads” language and why hallucinations happen.
Beyond opening a chat UI, they can describe tokens, training data and hallucinations, and judge when outputs need human checking. Understanding stops generative AI being treated as an always-correct answer machine.
02
Four-part prompts + Vibe Coding: natural language into Python projects.
Clarify role, task, constraints and examples, then drive coding with natural language: AI drafts, students review and revise. “Produce” means runnable, demonstrable work—not a pile of unsorted prompts.
03
Projects, competitions, school showcases—and ethics—leave with the student.
After class, students take a personal or group project plus basic awareness of bias, privacy and copyright. Work can map to school STEAM Day or related showcases, extending learning beyond the classroom.
Four units: foundations → Prompt → Vibe Coding → project pathways
VibeCode HK is tailored for Hong Kong schools. The same approach can flex by lesson time, student level and project focus.
Interactive demos
Light demos that mirror the “see it instantly” classroom feel—all simulated in the browser. No API keys required.
Instant platform output · Students do not need their own API keys by default
Related · Prompt Engineering →
Pick a preset or edit the prompt, then Run to see a simulated reply—practise making requirements clear.
Presets
Demo only: the platform does not ask students for API keys; real classes use school / platform backends.
Simulated AI reply
—
Why learn now
01
Students can ask and paste answers, yet cannot say what an LLM is or why it errs. If class stops at “using the tool”, they face hallucinations by asking again instead of judging reliability. Schools need explainable foundations and output checking—not only chat logs.
02
One error and they stop. Many never turn a natural-language idea into a finished piece. Traditional intros memorise syntax before results appear; Vibe Coding clarifies the goal first, then verifies with a Python project while students practise reading and fixing code.
03
Tools without a curriculum. Teachers struggle to take over; parents cannot see outcomes. Schools often buy software or accounts yet lack a path from foundations to projects, so progress and showcases stall. They need unit design that coordinators and IT teachers can hand on.
04
No portfolio, no competition direction—only chat logs when class ends. Parents and schools ask whether students can demo a piece of work for school STEAM Day or related showcases. Without projects and ethics, learning rarely extends beyond the classroom.
Course outline
A unit teaser below. Full detail, paragraphs and points are on the course page.
01
Why everyone talks about LLMs—and how they actually “read” language.
02
Ask in a way that actually helps—verifiable and reusable.
03
Turn ideas into Python projects with natural language—not rote syntax.
04
Take work away for competitions, school showcases, further study and ethics.
After the course
After four units, students should explain concepts, write usable prompts, demo work, and carry ethics awareness into showcases or school follow-on.
Describe large language models and natural language processing in their own words, separate what they can and cannot do, understand why hallucinations appear, and check outputs as habit.
Use the four-part method to turn fuzzy ideas into checkable instructions for writing, problem-solving and coding help; follow up and refine instead of quitting after one weak answer.
State needs in natural language, finish a runnable Python mini-project, and read or fix code. Students own the work rather than pasting model output blindly.
Complete a personal or group LLM application project aligned to showcase or competition directions, with awareness of bias, privacy, copyright and responsible use for school and after-class settings.
For secondary students, upper primary, and schools—including international schools—planning an AI course.
Tailor sessions, level and projects for your school
Phone 2612 5789 · Fax 2612 5787
WhatsApp 5744 5000 · Email info@vibecode.hk
Website vibecode.hk