The biggest difference between school-leavers who get good first jobs and ones who do not is having something real to show. Not a list of subjects. Not a self-described skill. A working thing that you made.
In 2026, the cheapest way to make a real thing is with AI. The portfolio below is three small projects: a script, a one-page site, a data analysis. Each takes 4 to 12 hours of focused work and proves a different skill.
You do not need to be technical. You do not need to spend money. You do need to actually finish.
Project 1: A useful script you wrote with AI help
What you build: a script that does one boring thing automatically. Examples:
- Reads a folder of PDF receipts and pulls out the total amount, date and supplier into a spreadsheet.
- Renames every photo on your phone by date taken.
- Scrapes a news site's headlines once a day and emails them to you.
- Generates 30 flashcards from a set of class notes you paste in.
Tooling:
- A free OpenAI or Claude account for the AI assistant.
- A free Python or JavaScript install, or Google Colab in the browser.
- About 4 to 8 hours.
The build process:
- Pick the problem. Something you actually have, not something you imagine.
- Ask AI to outline the steps. Read the outline. Ask why for any step you do not understand.
- Ask AI to write the script step by step. Run each step and check the output.
- When it breaks, paste the error back to AI. Read its diagnosis. Try the fix. Repeat.
- When it works on one file, test it on five. Then ten.
- Write a 100-word README explaining what it does and how to run it.
- Put it on GitHub (free) or a public Google Drive folder.
What you have proven: you can spec a problem, work with AI to solve it, debug iteratively and ship something that runs.
Project 2: A one-page website with AI-generated content
What you build: a personal one-page site that is genuinely yours. Examples:
- A study-plan generator: paste your subjects and get a printable weekly plan.
- A school-leaver cost-of-living calculator for moving out.
- A landing page for a school-formal carpooling roster.
- A simple quiz on a topic you like (e.g. Year 12 chemistry, AFL trivia).
Tooling:
- AI assistant for code and content.
- A free hosting account: Vercel, Netlify, GitHub Pages or Cloudflare Pages.
- A domain (optional, $15 a year if you want).
- About 6 to 12 hours.
The build process:
- Decide the one thing the page does. Just one.
- Ask AI to scaffold a single-page HTML/CSS/JS app that does it.
- Open the file in your browser. Make sure it works.
- Tweak the styling, the copy, the layout. Make it look like a thing a real person would visit.
- Deploy to Vercel or Netlify by dragging the folder into their dashboard.
- Share the link.
The trick: do not let AI write copy that sounds like AI. Rewrite every word the user reads in your own voice.
What you have proven: you can specify a small product, work with AI to build it, deploy something to the public web with a URL you can put on a resume, and care about the user experience.
Project 3: A small data analysis with AI
What you build: a short analysis of a dataset you find online, with a one-page write-up. Examples:
- ATAR cutoff trends across NSW universities over five years.
- AFL or NRL outcomes by margin and team form.
- ABS census data on housing affordability in your suburb.
- Spotify charts behaviour for the songs you listen to most.
Tooling:
- AI assistant for help with code and statistics.
- Google Colab or a Jupyter notebook for the work.
- A public dataset (data.gov.au, ABS, AustLII, Kaggle, the BoM, AFL Tables).
- About 6 to 10 hours.
The build process:
- Pick a dataset you can actually download in a usable format (CSV is your friend).
- Open it in Colab and let AI help you load it, clean it and explore it.
- Pick one specific question the data could answer. Just one.
- Compute the answer. Make a chart.
- Write a 250 to 500 word explanation: what the question was, what the answer is, what the data does and does not let you conclude. This is the part that matters most.
- Save the notebook publicly on GitHub or Colab.
The ethics rule: every claim in your write-up must be traceable to the data, not to AI's guess. If AI generates an interpretation, you must verify it from the actual numbers.
What you have proven: you can find data, work with AI to clean and analyse it, draw a careful conclusion and communicate it in writing. That is junior data analyst, junior actuary, junior consultant, junior policy adviser and junior researcher all in one skill set.
What to put on your CV
When you have all three done, your CV or LinkedIn gets a small section:
Projects, 2026:
- Script that reformats class notes into Anki-ready flashcards. github.com/yourname/notes-to-anki.
- One-page study planner at yourplanner.netlify.app, used by 40 classmates.
- Data analysis: how NSW ATAR cutoffs have shifted 2021-2025, github.com/yourname/atar-trends.
Tools: ChatGPT/Claude, Python, GitHub, Vercel.
That section is more useful to a recruiter than half a page of generic competencies.
What to say in an interview
You will likely be asked about a project. Have one 90-second answer ready per project:
- What it does in one sentence.
- What was hard about it (be honest, not modest, not boastful).
- What you used AI for and what you did yourself.
- What you would change if you did it again.
That answer signals everything employers want: you can finish things, you can talk about your own work, you used AI well, and you are self-aware.
The honesty principle
When you show this portfolio, you should be able to:
- Explain every line of code if asked.
- Defend every claim in your data write-up.
- Run the project live in front of someone.
If you cannot, AI did too much of the work. Go back and do the part you skipped. The point of the portfolio is not to fool anyone. It is to prove to yourself, first, that you can ship.