Teen Portfolio Sprint: 3 AI Mini‑Projects in 30 Days (Prompts, Rubrics, GitHub Tips)

Why a 30‑Day “Portfolio Sprint” Works (and What Parents Should Expect)

A strong teen portfolio doesn’t need a huge, complicated “AI app.” What colleges, internship coordinators, and competition judges really want is proof that your student can:

• Start a project with a clear goal
• Make small, steady progress
• Explain choices and results
• Share their work in a professional way

That’s why a sprint is perfect. In 30 days, your teen can build three small AI mini‑projects—each one highlighting a different skill—then publish them in a polished GitHub portfolio for teens.

This post gives you a 30 day AI project plan, plus ready-to-use prompts, simple rubrics, and GitHub tips that make the work look credible (even if your teen is a beginner). These are intentionally designed as machine learning projects for beginners students can finish without burning out.

What parents should expect:

• Time: ~30–45 minutes/day, 5 days/week (with one longer “build day” on weekends)
• Tools: a browser, a free GitHub account, and either a notebook tool (like Google Colab) or a simple coding environment
• Outcome: 3 project folders on GitHub + short write-ups + demos/screenshots

The 3 Mini‑Projects (Pick One Track: No‑Code, Light‑Code, or Code)

To make this practical, each mini‑project below includes three “tracks” so your teen can match their experience level.

Mini‑Project 1: Prompt Engineering “Evaluator” (AI Communication)

Goal: Show that your teen can design prompts, test them, and measure quality.

Options:

• No‑Code: Compare 10 prompt versions that summarize a news article at different reading levels.
• Light‑Code: Write a small script/notebook that stores prompts + outputs in a table.
• Code: Build a simple prompt test harness that scores outputs (clarity, accuracy, tone) and saves results.

What goes in the portfolio:

• A prompt library (10–15 prompts)
• Before/after examples
• A short rubric showing how they judged outputs

Mini‑Project 2: Beginner ML Classifier (AI Modeling)

Goal: Train a small model and explain what it learned.

Options:

• No‑Code: Use a beginner-friendly tool that trains a classifier from examples (e.g., text sentiment or image categories) and export results.
• Light‑Code: Use a notebook with a starter template (logistic regression for text or a simple image classifier).
• Code: Build a clean notebook with train/test split, evaluation, and a confusion matrix.

Beginner-friendly dataset ideas:

• Text: positive/negative movie reviews, spam vs. not spam, sports vs. politics headlines
• Images: cats vs. dogs (small subset), recyclable vs. trash (simple categories)

What goes in the portfolio:

• A short dataset description (where it came from, how many samples)
• Model performance (accuracy + 1–2 other metrics)
• A “what went wrong” section (misclassifications)

Mini‑Project 3: AI Mini‑App Demo (AI Product Thinking)

Goal: Package the work into something others can try or understand quickly.

Options:

• No‑Code: A slide/demo walkthrough showing a user story and results.
• Light‑Code: A simple web page that takes input and shows an AI output (even if it’s a mocked sample).
• Code: A small app (Streamlit or similar) that runs the model/prompt workflow with a clean UI.

What goes in the portfolio:

• A short demo video or GIF
• A “user story” (who it helps, what problem it solves)
• Clear instructions to run it

These three together create a well-rounded set of ai projects for high school students portfolio reviewers love: communication, modeling basics, and product presentation.

The 30‑Day Plan (Weekly Milestones + What to Produce)

A sprint needs structure. Here’s a realistic plan that fits school schedules.

Week	Focus	Deliverables by Sunday	Parent Check-In (10 minutes)
1	Choose topics + set up GitHub	GitHub account, 1 portfolio repo, project ideas written down	Ask: “Which project are you most excited to show someone?”
2	Mini‑Project 1 (Prompts)	Prompt library + scoring rubric + 10 test outputs	Ask: “What changed from prompt v1 to v10?”
3	Mini‑Project 2 (Beginner ML)	Dataset summary + baseline model + evaluation screenshot	Ask: “Where does the model fail and why?”
4	Mini‑Project 3 (Mini‑App) + polish	Demo + final READMEs + pinned repos	Ask: “Could a stranger run this in 5 minutes?”

Daily rhythm (keep it simple):

• Mon–Thu (30–45 min): build one small piece
• Fri (20 min): write 5–8 lines in the README: what you did, what you learned
• Weekend (60–90 min): integrate + test + screenshot

A key tip: the portfolio “wins” aren’t just the code—they’re the explanations. Most teens skip that. Don’t.

Copy‑Paste Prompts + Simple Rubrics (So the Projects Look Professional)

These templates help your teen produce work that reads like a real project.

Prompt Pack for Mini‑Project 1 (Evaluator)

Have your teen pick one content source (an article, a chapter summary, a science topic) and test prompt variants.

Prompt A (Baseline Summary):

• “Summarize the text in 5 bullet points. Keep it factual.”

Prompt B (Audience Shift):

• “Summarize the text for a 9th grader. Use simple vocabulary and short sentences.”

Prompt C (Structure Control):

• “Create a summary with: 1 sentence overview, 3 key facts, and 2 important terms with definitions.”

Prompt D (Bias/Verification):

• “List 3 claims from the text and explain what evidence would be needed to verify each claim.”

Prompt E (Self‑Critique):

• “Write a summary, then critique your own summary for missing details, unclear wording, or possible inaccuracies. Provide an improved version.”

What makes this portfolio-ready: store each prompt version, output, and score in a simple table (even a spreadsheet screenshot is fine).

Scoring Rubric (Works for Prompts, Models, and Apps)

Use a 1–4 scale. The goal is to show the teen can evaluate quality, not just generate results.

• Clarity: Is it easy to understand?
• Correctness: Are there factual errors?
• Completeness: Does it cover the key points?
• Consistency: Does it follow the requested format/tone?

Your teen can add one project-specific category:

• For ML: “Handles edge cases”
• For an app: “User experience”

Mini‑Project 2 ML Rubric (Beginner-Friendly)

• Data Handling (1–4): Clean split, no leakage, clear labels
• Evaluation (1–4): Uses accuracy + confusion matrix (or precision/recall)
• Interpretation (1–4): Explains mistakes with 3 concrete examples
• Reproducibility (1–4): Notebook runs top-to-bottom without errors

Mini‑Project 3 Mini‑App Rubric

• Demo Quality (1–4): Video/GIF shows the full flow
• Instructions (1–4): Clear “How to Run” steps
• Design (1–4): Simple, readable interface
• Responsibility (1–4): Mentions limitations and safe use

Parents: this rubric is your secret weapon. It turns “my teen played with AI” into “my teen followed an engineering process.”

GitHub Tips That Make a Teen Portfolio Look Legit (Even for Beginners)

A great github portfolio for teens isn’t about fancy code. It’s about clear organization and proof of work.

Here’s a clean setup:

• Create one main repo: teen-ai-portfolio
• Inside it, add folders:
  • project-1-prompt-evaluator/
  • project-2-ml-classifier/
  • project-3-mini-app/
• Add a simple README.md at the root with links to each project

README Template (Copy/Paste)

Each project README should answer:

• What it does: 2–3 sentences
• Why it matters: 1 sentence (the “so what?”)
• How it works: bullets (inputs → steps → outputs)
• Results: 1–2 screenshots + 2–3 metrics or examples
• Limitations: what it can’t do yet
• Next improvements: 3 bullet ideas

Commit Messages Teens Can Actually Use

A portfolio looks stronger when it shows progress over time.

Good commit examples:

• add baseline prompts and scoring sheet
• train first model + save confusion matrix
• fix data split bug and re-run evaluation
• add demo gif and update run instructions

Avoid:

• stuff
• final final
• idk

Pin Repos and Add a One‑Line Bio

On the teen’s GitHub profile:

• Pin the portfolio repo
• Add a bio like: “High school student building beginner AI projects (prompts, ML, mini-apps).”
• Add a profile README if they’re motivated (optional, but nice)

Safety Note for Families

Remind your teen not to publish:

• Full name, phone number, school address
• Personal photos in datasets
• Private API keys (use environment variables or placeholders)

Next Steps: Start the Sprint This Weekend

If you want your teen to finish, the first 48 hours matter. Do these three steps:

• Pick a theme: sports, music, climate, study tips, local community—anything they care about
• Choose the track: no‑code, light‑code, or code (the best choice is the one they’ll complete)
• Create the portfolio repo + first README: even a placeholder builds momentum

A simple “Weekend Kickoff” checklist:

• Create GitHub account
• Make teen-ai-portfolio repo
• Add three project folders
• Write 3 sentences per project: goal + audience + what “done” looks like

If your teen completes these three mini‑projects in 30 days, they’ll have something rare: a portfolio that proves follow‑through. That’s the difference between “interested in AI” and “ready for real opportunities.”