AI & Robotics Youth FoundationDonate
Menu

Stage 02 of 04 · Builder

AI Builders Studio

Build with LLMs, not just chat with them.

AI · Coding

Ages
12 – 18
Duration
10 weeks
Prereq
C9 Python for Builders strongly recommended.

What students learn

How to build with LLMs and other AI tools — how to prompt them well, chain them, ground them with real data, and ship a small tool. Includes a serious AI ethics and safety thread.

Prompt-engineering instinct. A working API-integration pattern. The reflex to ask “who could this harm?” before shipping. Concrete experience evaluating AI output, not just trusting it.

Module-by-module

  1. What LLMs are

    Token prediction, context windows, failure modes. Annotated demo session.

  2. Prompt engineering basics

    Clarity, structure, examples, role prompting. 5 well-crafted prompts.

  3. Prompt engineering advanced

    Chain-of-thought, structured output, evaluation. Graded portfolio.

  4. Calling an API

    HTTP requests to an LLM, parse JSON. First programmatic call.

  5. Building with the API

    Streamlit/Flask wrapper around LLM. First AI mini-app.

  6. Grounding & RAG (light)

    Give the model real data; basic retrieval. Tool that answers questions about a real document.

  7. Safety & guardrails

    Refuse, redirect, sanitize. Safety case study + a guardrail in your tool.

  8. AI ethics in the wild

    Bias, privacy, environmental cost, who gets harmed. Ethics review of your own app.

  9. Final project, part 1

    Pick a real user. Working v1.

  10. Final project, part 2

    Test with real users, fix what’s broken. v2 + reflections.

What they make

A small AI-powered tool (chatbot, summarizer, classifier, or assistant) tested with at least 3 real users.

Walk away with

Prompt-engineering instinct. A working API integration pattern. An ethics reflex.

Free resources we recommend

  1. Microsoft GenAI for Beginners

    21 lessons, end-to-end LLM curriculum. Free GitHub Models endpoint included.

    Microsoft, MIT-licensed, 70k+ GitHub stars.

    Lessons + code samples + videos

  2. Hugging Face Learn

    Best open-weights perspective — teaches the right mental model.

    Hugging Face engineers; runs on free Colab.

    Written chapters + runnable notebooks

  3. Anthropic Prompt Engineering Tutorial

    9 chapters, self-grading exercises.

    Anthropic (makers of Claude); MIT-licensed.

    Jupyter notebook or Google Sheets

Honorable mentions

Next step

Innovator project that uses AI.

Back to the course catalog