From a working project to a portfolio asset you can show the world.
Shape: 20 min theory + 100 min practical · You'll need: the Week 5 Colab notebook + your project from Weeks 3–4
The shape of today
Polish, publish, present
2 halves
Theory (20 min): how agentic projects open real doors — résumés, apps, internships.
Practical (100 min): polish the code, publish to Kaggle, then pitch it live.
Ends with a Presentation Slam — 3-minute pitches, real feedback, awards.
1
Part 1 · Theory
Pathways to opportunities
Concept · Framing your work
Turn "I built a thing" into a story
Lead with the problem — "Parents can't compare summer camps easily, so I built…"
Name the agent skills — tool integration, planning, working with real data.
Show the artifact — a clickable Kaggle link beats any claim.
📄 Résumé line vs. college essay
Résumé: one crisp bullet with a link. Essay: the story of what broke, what you tried, what you learned. Same project, two lengths.
Concept · What companies want
The skills you've been building
🛠️ Tool integration
Wiring an LLM to functions, APIs, and data — the core of real agents.
🧩 Problem decomposition
Breaking a fuzzy goal into steps a system can actually execute.
🐛 Iterative debugging
Reading errors, forming a guess, testing it — the daily work of building.
🔎 Finding openings
Internships, research programs, hackathons, open-source issues.
🔑 Internship search tips
Search "high school AI internship" + your city · email a professor with your Kaggle link · enter a Kaggle "Getting Started" competition · contribute one small fix to an open-source repo.
⚡ Energizer · 10–12 min · on your feet
60-Second Hype
Pair up. Each person gets 60 seconds to pitch — partner listens, no interrupting.
Round 1: pitch a silly product (a self-petting cat robot). Round 2: pitch your real project.
Feedback in two lines: "I liked…" and "I wondered…" — then swap.
Printable prompt cards in Coach HQ → Week 5 energizer
2
Part 2 · Practical
Polish, publish, present
Practical · Station 1
Production polish
Comment the "why," not the "what" — future-you and your reader will thank you.
One clear README cell at the top: what it does, how to run it, what to expect.
Peer test: a partner runs your notebook top-to-bottom on a fresh copy.
# Give your agent a friendly face — one Gradio cellimport gradio as gr
gr.Interface(
fn=run_agent, # your agent function
inputs="text", # the user's question
outputs="text", # the agent's answer
title="Camp Finder Agent"
).launch() # shareable live demo link
Note
No time for a UI? A clean CLI loop (input() → agent → print()) counts as polish too. Pick one and finish it.
Practical · Station 2
Publish to Kaggle
Colab → Kaggle: download .ipynb, then New Notebook → File → Import on Kaggle.
Markdown storytelling: title, the problem, how the agent works, results.
Reproducibility: "Run All" must work top-to-bottom with no missing cells.
# A markdown cell tells the story around your code
## 🏕️ Camp Finder Agent
**Problem:** comparing summer camps by hand is slow.
**How it works:** the agent calls a search tool, then
summarizes the top matches into a short table.
**Result:** ask one question, get a ranked answer.
🔗 Deliverable
Set the notebook to Public and grab the link. That URL is your portfolio piece — paste it into the shared template.
Practical · Station 3
Portfolio Presentation Slam 🎤
1 · Problem + design
What you set out to solve and how your agent is put together.
2 · Live demo
Run it. Ask it a real question. Let the room see it work.
3 · Key learning
The hardest bug or biggest surprise — and what it taught you.
🗣 Feedback frame
After each pitch, two hands: one "I liked…" and one "I wondered…" Same frame as the energizer — keep it warm and fast.
🏆 Awards: Most Creative · Best Execution · Clearest Explanation
Week 5 · Finale
You have a portfolio 🎉
✅ A working Colab project — a real agent that runs end-to-end.
✅ A public Kaggle notebook — a link you can put on any application.
✅ A 3-minute pitch — you can talk about your work with confidence.
✅ A portfolio page — everything in one place, ready to share.
🗺️ Back to the roadmap
Talk → Agents → Data → Build → Ship. Five weeks, from "AI answers" to "I built and shipped an AI agent." Keep going — pick your next problem.
Next step → Enter one Kaggle competition · email your notebook link to someone who'd find it cool.
AI Trailblazers · Week 5 — Polish & Present · press S for coach notes