29 ChatGPT prompts for students to turn coursework into portfolio-ready projects
Hook: stop uploading PDFs that no recruiter will open
If you’re a college student in 2026, here’s the cold truth: nobody’s hiring you for the PDF of your group project. Employers hiring the Class of 2026 are looking for evidence - of teamwork, problem-solving, and communication - and NACE’s 2026 Job Outlook Spring Update found that 85% of employers screen for teamwork on a resume. A homework-grade write-up with no story, no metric, and no demo just doesn’t pass that screen.
That’s why I wrote this. Below is a 4-stage Course-to-Project Pipeline plus 29 ChatGPT prompts for student coursework to portfolio project work that actually get you interviews. Each prompt is multi-line, ready to paste, and follows a structure: purpose, the prompt itself, an example output, and a pro tip.
“85% of employers look for teamwork on the resume, and 13.3% of entry-level jobs now require AI skills.” - NACE Job Outlook 2026 Spring Update, April 2026
The goal isn’t to make you look busier. The goal is to make your three strongest pieces of coursework look like real, shipped work a hiring manager can scan in 60 seconds.
Quick answer / TL;DR
- Pick 3 projects, not 12. Recruiters will only click 3, and NACE’s 2026 Spring Update shows employers value quality of evidence over volume.
- Run every project through the 4-stage pipeline: PICK → SHAPE → SHIP → TELL. It’s a student version of the Design Council’s Double Diamond - discover/define on the left, develop/deliver on the right.
- Always include a “Built with” line. Per NACE’s April 20, 2026 report, 13.3% of entry-level jobs now require AI skills. Show yours.
- The 4 L’s of a Student Portfolio (Listen, Link, Learn, Lead) - NACE’s framework - should be visible on every case study page.
- Use GitHub Education’s free tools. GitHub Education gives 5 million verified students Copilot, Codespaces, and a student pack that powers the whole pipeline at zero cost.
Why your degree projects read like homework (and what 2026 data says)
Most student portfolios are basically a Dropbox folder. Mine was. The reason is that coursework optimizes for a rubric - citations, page count, problem set answers. It doesn’t optimize for a 60-second skim from a tired recruiter on a Tuesday afternoon.
The 2026 numbers make the stakes clearer:
- 5.6% more new college graduate hires projected for the Class of 2026, per NACE’s Spring 2026 Update. Hiring is up - but it’s concentrated in Information, Engineering Services, and Misc. Professional Services.
- 13.3% of entry-level job posts now require AI skills, and the share is “nearly triple” the Fall 2025 number, per NACE’s April 20, 2026 report. 28% of employers are seeking early talent with AI use; nearly 60% assign interns AI-powered projects.
- 70% of employers now use skills-based hiring, also from NACE 2026. They want examples, not declarations.
- 45% of employers rate the 2026 grad job market “fair” - the same word they used in 2021 when hiring was flat. The bar is higher than the optimism suggests.
Translation: more openings, more competition, more AI fluency expected. Your portfolio can’t be a polite reading list. It needs to show a problem you cared about, a tool stack you used (including AI), a number you moved, and a story a human can repeat in a 15-minute interview.
The other reason portfolios fail: students wait until senior year. The fix is the pipeline below.
The 4-stage Course-to-Project Pipeline
A framework I wish I had as a junior. Think of it as a student port of the Design Council’s Double Diamond, tightened for semester deadlines. Each stage has a single, shippable output.
Stage 1 - PICK (the “Discover/Define” diamond)
Goal: pick 3 projects from your transcript, not 12. Output: a one-page Project Brief with a real audience, a real constraint, and a real metric.
The 4 L’s of a Student Portfolio, defined by NACE for portfolio design: Listen to a stakeholder, Link to a job family, Learn a tool or concept, Lead a moment of ownership. Every pick should give you at least 3 of the 4.
Stage 2 - SHAPE (the “Develop” diamond)
Goal: turn the brief into a buildable plan. Output: a problem statement, a tools list, a wireframe or system diagram, and a “Definition of Done.” This is where the 29 prompts start earning their keep.
Stage 3 - SHIP (the “Deliver” diamond)
Goal: get the thing live. Output: a public GitHub repo (use GitHub Education’s free Student Developer Pack for hosting + credits), a deployed link, a 90-second Loom demo, and a clean README.
Stage 4 - TELL (the iteration loop)
Goal: make the work findable. Output: a case study page, a LinkedIn post, a 4-bullet resume entry, and a STAR-formatted interview story. Per NACE’s resume skills report, employers want examples - and STAR is the cleanest way to bank them.
Pull quote: “It is not enough for candidates to list their skills: employers want to see examples.” - Kevin Gray, NACE, April 20, 2026
The rest of this article gives you the 29 prompts to run inside that pipeline, in pipeline order.
Project-picker prompts (1–4)
These four prompts run at Stage 1 (PICK). Their job is to convert a transcript into a ranked short-list of 3 portfolio-grade projects.
Prompt 1 - The Transcript Sweep
Purpose: Quickly turn a list of every course, lab, and group project you’ve done into a ranked short-list of “portfolio-worthy” picks, scored against the 4 L’s of a Student Portfolio (Listen, Link, Learn, Lead).
Prompt:
You are a portfolio coach for a college student applying to entry-level
roles in [TARGET INDUSTRY, e.g., software engineering, data analysis,
UX design, product marketing].
Below is my list of every course, lab, capstone, internship project,
hackathon, club project, and side project I've worked on in the last
24 months. Each line is roughly: [Course / Project name] - [one-line
description] - [tools used].
[List your projects here, one per line.]
Score each project from 1–5 on each of NACE's "4 L's of a Student
Portfolio":
- Listen: Did I observe or interview a real stakeholder/user?
- Link: Does it map to a real job family or industry problem?
- Learn: Did I pick up a tool, model, or method I can re-use?
- Lead: Did I own a decision, deadline, or piece of the work?
Then, in a table, return:
1. Project name
2. Score per L (1–5)
3. Total score
4. One-sentence "what is the portfolio version of this?"
5. Effort to turn it portfolio-ready (Low / Medium / High)
Finish with a top-3 ranked recommendation and a 2-sentence rationale
for each.Example output (abridged):
Project: ECON 312 Pricing Lab - Listen 2, Link 4, Learn 3, Lead 3 → 12/20 - Effort: Low. The portfolio version is “Built a 3-variable pricing model in Python that recomputed unit economics for a local coffee shop and predicted a 6% margin lift.”
Pro tip: Don’t be precious. If a project doesn’t have at least 3 of the 4 L’s, archive it. The “Effort to turn it portfolio-ready” column is your real filter.
Prompt 2 - Audience Reality Check
Purpose: Force every shortlisted project to name a real human it would help, so you stop writing for “anyone interested in.”
Prompt:
I'm picking 3 portfolio projects for [TARGET ROLE, e.g., junior data
analyst, associate product designer, marketing coordinator].
For each project below, do the following:
1. Name one specific person who would read this case study
(title, company type, and what they're hiring for).
2. Write the first sentence of the case study that names a
problem *that person* cares about. Boring beats clever.
3. List 3 questions that person would ask in a 15-min interview
about this project.
4. Suggest one screenshot, chart, or artifact that would answer
question #1 instantly.
Projects:
- [Project A - one-paragraph description]
- [Project B - one-paragraph description]
- [Project C - one-paragraph description]Example output (abridged):
“Sara, a 28-year-old marketing ops lead at a mid-size DTC skincare brand, would open this case study if it promised a 4-step playbook for cutting her weekly reporting time by half. Her first question: ‘How do you know the time savings wasn’t just because you stopped doing X?’”
Pro tip: Re-use the “name a person” trick on your resume. One named-audience bullet on a resume is worth five generic ones. This is also the fastest way to learn Jobs-To-Be-Done framing without reading Clayton Christensen.
Prompt 3 - The 60-Second Test
Purpose: Simulate the skim a recruiter gives each portfolio piece and force a brutal edit pass.
Prompt:
You are a recruiter at a 200-person Series B SaaS company. You have
60 seconds per portfolio piece. I will paste a portfolio case study
below.
Act like a real recruiter. Tell me:
1. The first thing your eye lands on (and what it tells you).
2. The single line that makes you keep reading, and the line that
makes you bounce.
3. The one question the case study *fails* to answer for you.
4. A specific rewrite of the opening paragraph (≤ 80 words) that
would hold you past 60 seconds.
5. The "Built with" / tools line that should appear in the
first screen (because the role mentions [STACK]).
Case study:
[PASTE YOUR DRAFT HERE]Example output (abridged):
“My eye lands on the title, which currently reads ‘Marketing Plan Project.’ Boring. I’d bounce at the abstract block. Rewrite: ‘In 11 weeks, I helped a 2-person student team rebuild [Campus Org]‘s Instagram strategy and grew weekly engagement 38% week-over-week. Here’s the playbook.’”
Pro tip: Paste every draft through this. The first 3 runs feel mean. The 4th teaches you the pattern and you stop needing it.
Prompt 4 - Star-Seed Builder
Purpose: Convert each project into a bankable STAR story (Situation, Task, Action, Result) so your next interview doesn’t start from scratch.
Prompt:
I'm prepping behavioral interviews for [TARGET ROLE].
Below are 3 projects. For each, generate 3 distinct STAR stories in
a single Markdown table with columns:
Situation | Task | Action | Result | Quantified Result | NACE
Competency Demonstrated (Teamwork / Problem-Solving / Communication
/ Critical Thinking / Leadership / Technology / Professionalism /
Career & Self-Development)
Projects:
1. [Project A - short description]
2. [Project B - short description]
3. [Project C - short description]
After the table, list the 3 "I don't have a story for this yet"
competencies I should backfill before my next interview.Example output (abridged):
Project Situation Task Action Result Comp. Pricing Lab ECON 312 partner was a 4-person team. Forecast Q3 unit margin after a 6% supplier hike. Built a Python pricing model; ran 3 scenarios in 4 hrs. Predicted 4.2% margin loss; team changed SKUs and avoided it. Critical Thinking
Pro tip: This pairs with NACE’s 8 Career Readiness Competencies (NACE 2025 definition update). The “I don’t have a story” list is your interview prep homework.
Context & audience prompts (5–9)
Stage 2 (SHAPE) starts here. These prompts give your project a real audience, a real constraint set, and a real story spine.
Prompt 5 - The Doubled-Diamond Brief
Purpose: Force every project through a Discover → Define → Develop → Deliver pass, modeled on the Design Council’s Double Diamond, before you write a single line of code.
Prompt:
Act as a senior product designer. I'm a [YEAR + MAJOR] student working
on: [PROJECT NAME, 1-2 sentence description].
Walk this project through the Design Council's Double Diamond in
plain language, no design jargon.
- DISCOVER (divergent, ~150 words): What do I need to find out about
the user, the problem space, and the constraints? Give me 5
research questions, plus 2 cheap, fast ways to answer them this
week.
- DEFINE (convergent, ~150 words): A 1-sentence problem statement,
a 1-sentence "how might we" question, and the 1 metric that
would prove this project worked.
- DEVELOP (divergent, ~150 words): 5 candidate directions
(low-fi), the assumptions each one is built on, and the
cheapest prototype to falsify each.
- DELIVER (convergent, ~150 words): The smallest possible thing
I can ship in 7 days, what "done" looks like, and the one
decision I'm explicitly NOT making yet.
Tone: candid, like a senior telling a junior to stop over-thinking.Example output (abridged):
DISCOVER: Who else has tried this? What 2 questions can I ask 5 of them this week? Start with the people who gave up.
Pro tip: Keep the four outputs in a brief.md at the root of your project repo. It becomes the spine of your case study and the easiest thing in the world for a recruiter to skim.
Prompt 6 - User Persona in One Page
Purpose: Make the audience for each project as concrete as a real person. Useful for UX, CS, marketing, even data science.
Prompt:
Write a one-page user persona for the primary user of this project:
[PROJECT NAME + 2-sentence description].
Constraints:
- The persona must be 1 named person, 22–40 years old, with a
real-sounding job title.
- Include 3 jobs-to-be-done (functional, emotional, social).
- Include 3 frustrations that any solution in this space must
avoid.
- Include a 2-sentence "Quote" in their voice.
- Include a "Day in the life" 5-bullet timeline that ends with
the moment they would discover the project I'm building.
Output as Markdown, max 350 words.Example output (abridged):
Maya, 26, Operations Analyst at a 40-person health-tech startup. “I just want one report my CEO will actually read.” Day in the life ends at 4:47 PM when her boss DMs “got 2 min?” and she has 0 prepared views.
Pro tip: “Persona” doesn’t mean “fake.” Pull a real person you’ve talked to and fictionalize the name + company. The texture stays real, the privacy stays safe.
Prompt 7 - Problem Statement Refinery
Purpose: Kill the vague “I made a thing” energy and replace it with a one-sentence problem a recruiter can repeat in their head.
Prompt:
Below are 3 draft problem statements for my portfolio project.
For each, do the following:
1. Identify the 3 weakest words (jargon, hedge, or fluff).
2. Rewrite the statement using this exact template:
"[Specific user] struggles with [specific problem] when
[specific context], which causes [specific consequence].
I built [specific intervention] to [specific outcome]."
3. Score the rewrite on a 1–5 "Could a stranger repeat this
back?" rubric.
Drafts:
- [Draft 1]
- [Draft 2]
- [Draft 3]Example output (abridged):
Rewrite: “First-year community college students struggle to map their transfer credits to a 4-year CS pathway within the first advising appointment, which causes 31% of them to lose 1–2 semesters. I built a credit-mapping web app that cuts that advising call from 40 min to 12 min.”
Pro tip: If you can’t fill the template with real numbers, your project isn’t done yet. Go back and measure something.
Prompt 8 - Stakeholder Map
Purpose: Make your project feel like it was built with people, not just at people. Recruiters read this.
Prompt:
List every stakeholder group for this project, who within each group
would care about this, and what each one would want the case study
to mention.
For each stakeholder, give me:
- Their role
- The one thing they would Google after reading the case study
- The single line in my case study that would make them reply
Group categories: users, decision-makers, technical reviewers,
external partners, internal skeptics.
Project: [PROJECT NAME + 2-SENTENCE DESCRIPTION]Example output (abridged):
Technical reviewer (a senior eng hiring for the team). They’d Google “how the API auth works.” The line that wins them: “Auth uses Clerk with PKCE flow; refresh tokens never touch client storage.”
Pro tip: This map doubles as your “who to send this to on LinkedIn” list. Send the case study to one person from each row.
Prompt 9 - Constraint & Trade-off Memo
Purpose: Show that you made real decisions under real limits. This is the single biggest gap between homework and portfolio.
Prompt:
Write a 1-page "Constraints & Trade-offs" memo for this project,
in the voice of an engineer shipping under deadline.
Sections:
- Hard constraints (deadline, data, compute, team size, budget)
- Soft constraints (style, scope creep, my own skill gaps)
- 3 trade-offs I made and what I gave up by making each
- 1 thing I would do differently with 2 more weeks
- 1 thing I would NOT change even with infinite time
Project: [PROJECT NAME + 2-SENTENCE DESCRIPTION]
Length: ≤ 400 words.Example output (abridged):
Trade-off: I chose SQLite over Postgres for the demo so I could one-click deploy on Vercel. Gave up: multi-user write concurrency. With 2 more weeks, I’d add row-level security. I would not change: shipping the rough demo on day 7 over a perfect repo.
Pro tip: This memo, dropped at the bottom of your README, is the single most underused asset in student portfolios. It’s also the easiest thing for a hiring manager to ask you about in an interview.
Process & build prompts (10–14)
These run during Stage 2 (SHAPE) and Stage 3 (SHIP). They help you plan, scaffold, and build without sounding like a tutorial.
Prompt 10 - 7-Day Build Plan
Purpose: Convert the brief into a 7-day calendar with checkpoints you can defend in a standup.
Prompt:
Make me a 7-day build plan for this portfolio project, working
evenings and weekends (≈ 12 hours total). I'm a [YEAR + MAJOR]
student. Output as a Markdown table with columns: Day | Block |
Hours | Outcome | Commit message | Visible artifact.
Constraints:
- Every day must end with a public artifact (commit, deploy,
screenshot, or 30-sec Loom).
- The plan must include 1 "ugly day" where the goal is to ship
something I'd normally hide.
- Day 7 must end with a public case study draft.
Project: [PASTE YOUR BRIEF.MD HERE]Example output (abridged):
Day Block Outcome Commit Artifact 1 1 hr Problem statement, 5 user quotes docs: brief + research brief.md 2 2 hr Wireframe in Figma design: v1 wireframe Figma link 3 3 hr Replit skeleton + auth feat: auth + landing staging URL
Pro tip: Treat the “Visible artifact” column as a standup script. The recruiter-friendly version of this is the public commit log.
Prompt 11 - Tech Stack Sorter
Purpose: Stop over-engineering. Pick the smallest stack that ships, and defend the choice.
Prompt:
Suggest the minimum viable tech stack for this project:
[PROJECT DESCRIPTION + MUST-HAVE FEATURES].
Constraints:
- I want to deploy for free using [GitHub Education Student Pack
credits / Vercel free tier / Netlify free tier].
- I have [N] weeks and know [LIST OF LANGUAGES/TOOLS].
- The case study needs to show at least 1 AI-assisted feature.
Return:
1. Stack (frontend, backend, db, auth, hosting, AI piece)
2. Why this stack is smaller than the obvious one
3. What I'd give up by going simpler
4. The 2 Stack Overflow / GitHub issues I'm most likely to hit
in week 1
5. The fallback plan if [RISK] happensExample output (abridged):
Stack: Next.js (App Router) on Vercel, Supabase for auth + Postgres, OpenAI API for the AI feature, Loom for the demo. Why smaller than the obvious one: cuts out a custom backend. Give up: fine-grained rate limits.
Pro tip: Use the GitHub Education Student Pack to get free domain, hosting credits, and tools. GitHub Education serves 5M students - most of them aren’t using the credits.
Prompt 12 - README Skeleton Generator
Purpose: Generate a 2-minute README that a senior engineer would actually read.
Prompt:
Generate a README.md for this project, in the voice of a junior
engineer who respects the reader's time. Sections, in this order:
1. One-sentence project description
2. 30-second demo (link to Loom / GIF)
3. The problem, in 2 sentences
4. What I built, in 4 bullets with screenshots
5. How I built it (stack + key architecture choices)
6. What I'd do next, in 3 bullets
7. What I'd do differently, in 2 bullets (honesty wins)
8. "Built with" line listing all tools, including ChatGPT,
GitHub Copilot, and any AI feature
9. How to run locally (≤ 5 commands)
Project: [PASTE YOUR BRIEF.MD + TECH STACK]Example output (abridged):
Built with: Next.js 14, Supabase, OpenAI gpt-4o-mini, Vercel, Figma, ChatGPT (planning + refactor), GitHub Copilot (autocomplete), Loom (demo).
Pro tip: Per NACE’s April 20, 2026 AI skills report, 13.3% of entry-level jobs now require AI skills. Listing ChatGPT and Copilot in “Built with” is no longer optional.
Prompt 13 - Code Review Simulator
Purpose: Catch the dumb things before a senior engineer does.
Prompt:
Act as a senior engineer doing a 20-minute code review on my PR.
Here is the diff or repo link: [LINK OR PASTE].
Return:
1. The 3 things that would make you reject the PR (be specific
and unkind).
2. The 2 things that surprised you in a good way.
3. The single comment that would teach me the most.
4. The single comment that would annoy me.
5. A 5-line PR description rewrite that a senior would approve
without edits.Example output (abridged):
Reject: no error handling on the OpenAI call, secrets in client code, no test for the credit-mapping function. Annoy: “works on my machine” in the PR body.
Pro tip: Run every PR through this. It also doubles as a learning loop - keep a learned-from-review.md in your repo.
Prompt 14 - Loom Demo Script
Purpose: Write a 90-second Loom script that any teammate could record and that doubles as your case study’s hero video.
Prompt:
Write a 90-second demo script for a Loom recording. The audience
is a recruiter who has 90 seconds. Format as a Markdown table with
columns: Time | On screen | Voiceover | B-roll cue.
Rules:
- First 10 seconds must state the problem, the user, and the
outcome in plain English.
- Show 1 "ugly moment" - a real bug, a real recovery.
- End with a 1-sentence call to action ("Read the case study
here" / "Try the demo here").
Project: [PROJECT NAME + DEMO LINK]Example output (abridged):
| 0:00–0:10 | Cold open: blank dashboard | “If you’ve ever spent 40 minutes in an advising office trying to map transfer credits, this is for you.” | Quick cut to you on webcam | | 0:40 | Demo: paste transcript | “Here’s the moment it usually broke for me…” | Show the error |
Pro tip: Record with a script, but don’t read the script. Print it. Talk over it. The 2nd take is always better.
Outcome & metrics prompts (15–19)
These run at the end of Stage 3 (SHIP). The hardest part of any student project: making the result sound real. Per NACE’s 2026 skills report, employers want examples, not adjectives. Metrics are the difference.
Prompt 15 - Metric Mining
Purpose: Find 3–5 real, defensible numbers in a project you thought had none.
Prompt:
Below is a one-paragraph description of my student project:
[PASTE].
Mine it for every defensible metric. For each candidate metric,
return:
1. The metric
2. How I would have measured it
3. The honest baseline (or "no baseline - only post-measure")
4. A 1-sentence version of it that's safe to put on a resume
If a metric doesn't exist, say so. Don't invent numbers.Example output (abridged):
Metric: “Reduced advising call time from 40 min to 12 min.” Measurement: stopwatch on 3 simulated calls. Baseline: pre-app interview, average 38 min. Resume line: “Cut simulated advising call time 70% in a 5-week pilot with 6 mock users.”
Pro tip: “Simulated” is honest and still counts. Recruiters care that you measured something.
Prompt 16 - Outcome Narrative
Purpose: Turn 3 metrics into a tight 4-sentence story for the top of the case study.
Prompt:
I have these 3 metrics from my project:
1. [METRIC 1]
2. [METRIC 2]
3. [METRIC 3]
Write 3 different 4-sentence "outcome narratives" for the top of
my case study. Each one must:
- Sentence 1: Name the user and the problem.
- Sentence 2: Name what I built.
- Sentence 3: Quote 1 of the 3 metrics.
- Sentence 4: State what this proves the user (or I) can do next.
Tone options: data-driven / human-driven / systems-driven.Example output (abridged):
“For first-year transfer students, the first advising call is the one that decides whether they graduate. I built a credit-mapping web app to make that call shorter and clearer. In a 5-week pilot with 6 mock users, simulated advising time dropped 70% - from 40 to 12 minutes. That suggests a small, focused tool can move a stubborn institutional bottleneck.”
Pro tip: The “human-driven” version is the one to use when the metric is small. The “data-driven” version is the one to use when the metric is large. The “systems-driven” version is for senior interviewers.
Prompt 17 - Honest Failure Postmortem
Purpose: Write the “What I’d do differently” section that almost no one writes, and that almost every interviewer asks about.
Prompt:
Write a 200-word honest postmortem of this project. Tone: candid,
self-aware, no self-flagellation.
Sections:
- 1 decision I would NOT change
- 1 decision I would change
- 1 thing I learned about working with people, not just code
- 1 thing I learned about my own working style
- 1 sentence I'd say out loud if asked "what did you really
learn from this?"
Project: [PROJECT NAME + 2-SENTENCE DESCRIPTION]Example output (abridged):
I would not change shipping the rough demo on day 7. I would change waiting until week 3 to write tests. I learned that on a 3-person team, “I’ll do it tomorrow” becomes a 2-week delay. I learned I work best when I commit code in the morning and write in the evening.
Pro tip: This is the single highest-ROI section in the whole case study. Most students skip it. Hiring managers read it first.
Prompt 18 - Resume Bullet Rewriter
Purpose: Convert the case study into 3 resume bullets in the language hiring managers actually scan for.
Prompt:
Turn this case study into 3 resume bullets:
[PASTE CASE STUDY OR PROMPT 16 OUTPUT].
Rules:
- Each bullet ≤ 2 lines.
- First word is a strong verb.
- Each bullet must include a metric.
- Use NACE's 8 Career Readiness Competencies vocabulary where
natural: Teamwork, Problem-Solving, Communication, Critical
Thinking, Leadership, Technology, Professionalism, Career &
Self-Development.
- No "responsible for" or "helped with."
Return 3 versions:
A) Standard corporate resume
B) Skills-based resume (NACE 70% of employers, 2026)
C) 1-line version for LinkedInExample output (abridged):
A) Shipped a credit-mapping web app (Next.js, Supabase, OpenAI) that cut simulated advising call time 70% in a 5-week pilot with 6 mock users. B) Demonstrated Problem-Solving, Critical Thinking, and Technology competencies by designing and piloting a 1-developer tool that reduced a 40-minute institutional workflow to 12 minutes. C) Built a 1-person tool that cut advising call time 70% in a 5-week pilot.
Pro tip: Use the Skills-based version (B) for NACE-style skills-based applications. NACE found 70% of employers use skills-based hiring in 2026.
Prompt 19 - Case Study Skeleton
Purpose: Generate the full case study page from the work you’ve already done in the previous prompts.
Prompt:
Generate a 1,200-word case study page in Markdown, in this exact
order:
1. Title + 1-sentence summary
2. The user (1 paragraph, name + job + day-in-the-life)
3. The problem (1 paragraph, with the cost of not solving it)
4. The brief (3 bullets from Prompt 5)
5. The build (3 bullets + stack list + 1 architecture diagram)
6. The outcome (Prompt 16 + 3 metrics + 1 quote from a test
user)
7. The honest postmortem (Prompt 17)
8. The "Built with" line
9. The 1-line CTA
Constraints:
- Use the outputs of Prompts 5, 6, 9, 11, 12, 16, 17.
- Read like a senior wrote it for a junior, not a junior wrote
it for a senior.
- Include 2 inline screenshots placeholders labeled [Screenshot:
description].
Inputs:
[PASTE OUTPUTS OF PROMPTS 5, 6, 9, 11, 12, 16, 17]Example output (abridged):
The case study is too long to inline here, but it should fit on one scroll on a Notion page or a Carrd site, run ~7 min to skim, and ~25 min to read deeply.
Pro tip: Save this as a Notion template. Every new portfolio piece takes 90 minutes from blank to published using this skeleton.
Story & reflection prompts (20–24)
These run at Stage 4 (TELL). The work is shipped; now you have to make it sound like a story, not a status update. Per NACE 2026, employers want examples - and stories are the densest way to bank them.
Prompt 20 - The 4-Sentence Project Story
Purpose: Replace “I did a project for class” with a story that has a beginning, middle, and a cost.
Prompt:
I have a 5-minute project story that I keep telling in interviews,
but it always sounds like a list. Help me turn it into 4 sentences.
Inputs:
- Project: [NAME]
- User: [ONE SENTENCE]
- The biggest decision I made: [ONE SENTENCE]
- The metric that proves it worked: [ONE SENTENCE]
- The thing I would NOT have done without this project:
[ONE SENTENCE]
Write 3 versions of the 4-sentence story:
A) For a recruiter screen (90 seconds of airtime)
B) For a hiring manager interview (5 min of airtime)
C) For a portfolio site "About the project" page (read in 30 sec)
For each, label the sentences: Setup / Tension / Turn / Payoff.Example output (abridged):
A) “Transfer students lose 1–2 semesters to bad advising calls. Setup. I built a one-page tool to make that call shorter. Tension. Mid-build, I realized my ‘users’ were never going to log in twice. Turn. I cut auth entirely and went single-session. Payoff. Simulated advising time dropped 70%.”
Pro tip: Use Setup/Tension/Turn/Payoff as mental scaffolding in every interview answer. It’s a student-friendly port of STAR.
Prompt 21 - Lesson Extractor
Purpose: Pull transferable lessons from the project so your next interviewer doesn’t hear the same story twice.
Prompt:
From this project - [PROJECT NAME + 1-PARAGRAPH DESCRIPTION] - pull
out:
1. 3 transferable lessons (not project-specific)
2. 1 surprising lesson that I did not expect to learn
3. 1 question I now ask at the start of every new project
because of this experience
4. The 1 sentence I'd put on a personal "operating principles"
page
Tone: specific, not inspirational. Boring is fine.Example output (abridged):
- Talk to the user before drawing a wireframe. Measure something before declaring a win. Ship a thing that embarrasses you, then improve it.
- I learned that my best work happens in 90-min blocks, not 4-hour blocks.
- “What does ‘done’ look like on day 7?”
Pro tip: Save the 4 outputs as a “Lessons.md” file in your repo. Pull from it for cover letters, interviews, and “About” pages.
Prompt 22 - NACE Competency Mapper
Purpose: Map each project to the 8 NACE Career Readiness Competencies so you can talk to recruiters in their own language.
Prompt:
Map this project to NACE's 8 Career Readiness Competencies
(2025 update):
1. Career & Self-Development
2. Communication
3. Critical Thinking
4. Equity & Inclusion
5. Leadership
6. Professionalism
7. Teamwork
8. Technology
For each competency, return:
- Did this project demonstrate it? Y/N
- One specific example sentence from the case study
- One sentence I'd say in an interview to demonstrate it
If the answer is "no" for 3 or more, suggest a 1-sentence revision
to the project that would change that.Example output (abridged):
Teamwork: Y. Example: “Coordinated a 3-person student team across 2 time zones for 11 weeks.” Interview line: “I ran our weekly standup; we never missed a sprint.”
Pro tip: This is also the cleanest way to fill out a NACE-aligned skills-based application. 70% of employers use them in 2026.
Prompt 23 - Cover Letter Anchor
Purpose: Turn one portfolio piece into the spine of a 250-word cover letter, without sounding generic.
Prompt:
I am applying to [ROLE] at [COMPANY]. Here are 3 things from my
portfolio that match: [PASTE 3 BULLETS FROM PROMPT 18].
Write a 250-word cover letter paragraph that:
- Opens with the project, not "I am writing to apply"
- Names 1 specific thing about the company that the project
connects to
- Includes 1 metric
- Ends with a 1-sentence ask
Return 2 versions: a "match-the-job" version and a "show-the-fit"
version.Example output (abridged):
A) “When I built a credit-mapping tool for transfer students, the hardest part wasn’t the code - it was getting 6 mock users to sit through a 12-minute advising call. The same skill is the one I’d bring to [Company]‘s onboarding team: shorten the user’s first session without losing the parts that matter.”
Pro tip: “Show-the-fit” works better for stretch applications. “Match-the-job” works better for roles you check every box on.
Prompt 24 - The “What I Wish I Knew” Post
Purpose: Convert a portfolio piece into a 600-word public post that demonstrates judgment, not just output. Great for LinkedIn or your own site.
Prompt:
Write a 600-word public post in the style of a thoughtful
practitioner, titled "What I wish I knew before I built
[PROJECT NAME]".
Structure:
- Hook (1 sentence)
- The 3 things I wish I knew, ordered by how much they cost me
- The 1 thing I got right by accident
- The 1 question this raises that I don't have a clean answer to
- A 1-sentence invitation to reply (LinkedIn loves questions)
Tone: candid, lowercase, no hashtags, no "delve."Example output (abridged):
What I wish I knew before I built a credit-mapping tool for transfer students: that the hardest part of the project wasn’t the code, it was getting one busy advisor to spend 20 minutes with me. Two things I got wrong: I built the auth flow first, and I optimized for a metric my users didn’t care about. One thing I got right by accident: shipping the demo on day 7.
Pro tip: “Lowercase, no hashtags, no delve” is doing real work. It’s the difference between a portfolio piece that gets passed around and one that gets archived.
Distribution & site prompts (25–29)
The portfolio exists. Now make it findable. These prompts handle the GitHub Education repo, the public site, the social post, and the day-to-day updates.
Prompt 25 - Repo Polish Checklist
Purpose: Make your GitHub repo look like a real engineer’s, not a student’s.
Prompt:
Audit my GitHub repo for portfolio-readiness:
[REPO URL or PASTE FILE TREE].
Return a checklist with 3 columns: [x] Done | [ ] To do | Section.
Sections to check:
- README (per Prompt 12)
- LICENSE (default: MIT)
- .gitignore (no .env, no node_modules)
- CONTRIBUTING.md
- CODE_OF_CONDUCT.md (Contributor Covenant)
- GitHub Actions: a passing CI badge
- Issues templates
- One pinned "Highlights" issue
- Releases: at least 1 named release
- About box: description, website, topics
- "Built with" line in the repo description
End with the 3 highest-leverage fixes I can do in 30 min.Example output (abridged):
Top 3 fixes: (1) Add a 1-line repo description in the About box that says who this is for. (2) Pin a “Highlights” issue with screenshots. (3) Add a passing CI badge - even a “hello world” GitHub Action counts.
Pro tip: GitHub Education verifies 5 million students, and the Student Developer Pack includes free private repos, Codespaces hours, and a domain. Most students don’t use any of it.
Prompt 26 - Personal Site Brief
Purpose: Build a 1-page portfolio site you can ship in a weekend with Carrd, Framer, or Vercel - not a 6-page WordPress site you’ll never finish.
Prompt:
Write a brief for a 1-page portfolio site for a college student.
Sections (in this order, top to bottom):
1. Name + role + 1-sentence "I build X for Y" tagline
2. 3 portfolio cards (use the 3 projects from Prompt 4)
3. 1-paragraph "Currently"
4. 1-paragraph "Selected work" links to 2 case studies
5. 1-paragraph "Elsewhere" with LinkedIn + GitHub + email
6. Footer with last-updated date
Constraints:
- Total word count ≤ 350.
- All tool links go to live demos, not "coming soon."
- Must look good on a phone.
Also return:
- A 3-line CSS plan
- A Notion / Carrd / Framer section order
- The 1 line I'd put in the page <title> tag for SEOExample output (abridged):
Page title: “[Name] - credit-mapping tools for first-gen transfer students.” CSS plan: serif headline, system sans body, 1 accent color, max-width 640px.
Pro tip: Ship the 1-page version. Add the 2nd page in 6 months. Add the 3rd in 12. The bar is “exists and loads fast.”
Prompt 27 - LinkedIn Launch Post
Purpose: Convert one finished project into a LinkedIn post that doesn’t sound like a homework brag.
Prompt:
Write a 200-word LinkedIn post announcing a finished portfolio
project. Structure:
- Hook (1 line, lowercase, no emoji)
- The user, not me (2 lines)
- The metric, with the baseline (1 line)
- 1 thing that surprised me (1 line)
- 1 ask (1 line: feedback? intros? another user?)
Rules:
- No hashtags.
- No "I'm excited to announce."
- No AI tells ("delve," "tapestry," "navigate the landscape").
- Tag 2–3 people who would actually care.
- End with a question, not a CTA.
Project: [PROJECT NAME + 1-PARAGRAPH SUMMARY]Example output (abridged):
I built a credit-mapping tool because a transfer student told me she lost a semester to a 40-minute advising call. The pilot cut simulated advising time 70%. One thing I didn’t expect: the most useful feedback came from a financial aid officer, not a student. Looking for 2 more advisors to try a 12-min demo - DM me.
Pro tip: Ship the post the same day you ship the project. Half-life on LinkedIn posts is 18 hours. Half-life on your portfolio is forever.
Prompt 28 - Outreach DM
Purpose: Turn a finished case study into a warm, non-icky cold message to someone who should see it.
Prompt:
Write 2 versions of a 90-word DM to a [ROLE, e.g., product manager
at a fintech] about a project I just shipped. The DM must:
- Reference 1 specific thing they recently posted, shipped, or
said publicly
- Connect it to 1 specific decision in my case study
- Offer 1 thing (a 5-min call, a test user, a specific question)
- Be sendable without editing
Version A: to someone I don't know.
Version B: to a 2nd-degree connection.
Project: [PROJECT NAME + 1-PARAGRAPH SUMMARY]Example output (abridged):
A) “Hi [Name] - your post on cutting onboarding from 14 to 4 steps stuck with me. I just shipped a similar tool for transfer students; cut a 40-min advising call to 12. Would you be open to a 5-min critique? No pitch.”
Pro tip: Send 5 of these a week. 1 will reply. That 1 is your next interview lead.
Prompt 29 - 30-Day Update Loop
Purpose: Keep the portfolio alive after the launch. Per NACE’s 2026 data, 45% of employers rate the grad market “fair.” The bar is activity, not a one-time build.
Prompt:
Generate a 30-day content-and-update plan for a portfolio that
just went live. Working backwards from the launch.
For each week, return:
- 1 portfolio update (commit, fix, or add)
- 1 short post (LinkedIn, X, or a Discord you actually read)
- 1 outreach action (DM, cold email, or reply)
- 1 measurement (what I'll check at the end of the week)
Rules:
- No "post daily" advice. Once a week, deeply, beats daily noise.
- The plan must be sustainable for a full-time student.
- The plan must include 1 "ship a small improvement" week and
1 "kill a feature" week.
Project: [PROJECT NAME + LAUNCH DATE]Example output (abridged):
Week 1: add 1 user-requested feature; LinkedIn post on the design trade-off; DM 3 advisors for a 5-min call. Measurement: 3 advisor replies.
Pro tip: The “kill a feature” week is the one nobody does and the one that reads best in interviews. “I removed the dashboard and the user retention went up” is a story.
Comparison table: 29 prompts vs. project stage vs. output
| # | Prompt | Pipeline stage | Primary output | Best for |
|---|---|---|---|---|
| 1 | Transcript Sweep | PICK | Ranked project shortlist | Deciding what to build |
| 2 | Audience Reality Check | PICK | Named persona per project | Grounding in a real user |
| 3 | The 60-Second Test | PICK | Edited case-study opening | Recruiter skim survival |
| 4 | Star-Seed Builder | PICK | STAR story bank | Behavioral interviews |
| 5 | Doubled-Diamond Brief | SHAPE | 4-paragraph brief.md | Project spine |
| 6 | User Persona in One Page | SHAPE | 1-page persona | Audience clarity |
| 7 | Problem Statement Refinery | SHAPE | 1-sentence problem | Headline quality |
| 8 | Stakeholder Map | SHAPE | Stakeholder table | Distribution list |
| 9 | Constraint & Trade-off Memo | SHAPE | 1-page memo | Senior-eng credibility |
| 10 | 7-Day Build Plan | SHAPE | Calendar with commits | Avoiding scope creep |
| 11 | Tech Stack Sorter | SHAPE | Stack + trade-offs | Shipping faster |
| 12 | README Skeleton | SHIP | README.md | First-pass skim |
| 13 | Code Review Simulator | SHIP | Comment thread | Pre-merge QA |
| 14 | Loom Demo Script | SHIP | 90-sec demo video | Case study hero |
| 15 | Metric Mining | SHIP | 3–5 real metrics | Outcome honesty |
| 16 | Outcome Narrative | SHIP | 3 story openers | Case study lead |
| 17 | Honest Failure Postmortem | SHIP | 200-word postmortem | Interview depth |
| 18 | Resume Bullet Rewriter | TELL | 3 resume bullets | Apply, screen, LinkedIn |
| 19 | Case Study Skeleton | TELL | Full case study page | Portfolio site |
| 20 | 4-Sentence Project Story | TELL | Story in 4 sentences | Interview storytelling |
| 21 | Lesson Extractor | TELL | 4 transferable lessons | Cover letters |
| 22 | NACE Competency Mapper | TELL | Competency map | Skills-based applications |
| 23 | Cover Letter Anchor | TELL | 250-word paragraph | Job applications |
| 24 | ”What I Wish I Knew” Post | TELL | 600-word public post | LinkedIn / blog |
| 25 | Repo Polish Checklist | TELL | Audit checklist | GitHub credibility |
| 26 | Personal Site Brief | TELL | 1-page site brief | Carrd / Framer / Vercel |
| 27 | LinkedIn Launch Post | TELL | 200-word post | Public launch |
| 28 | Outreach DM | TELL | 2 DM templates | Warm intros |
| 29 | 30-Day Update Loop | TELL | Weekly rhythm | Keeping it alive |
People Also Ask: 9 questions students ask about turning coursework into portfolio projects
1. How do I turn a class assignment into a portfolio project without it looking like homework?
Pick 1 audience, 1 metric, and 1 “what would I do with 2 more weeks” line. That’s the minimum that separates a class assignment from a portfolio project. Run your draft through Prompt 3 (the 60-Second Test) and Prompt 7 (the Problem Statement Refinery). If the rewrite doesn’t read like a real product, you’re not done.
2. How many portfolio projects do I need?
3 is the magic number. NACE’s 2026 data shows employers scan resumes for evidence of skills demonstrated, not project volume. Three deep, well-told projects beat nine shallow ones every time.
3. What is the 4-stage Course-to-Project Pipeline?
The pipeline in this article: PICK → SHAPE → SHIP → TELL. It’s a student-friendly port of the Design Council’s Double Diamond, tightened for semester deadlines. Each stage has a single shippable output: a brief, a plan, a deployed project, a public case study.
4. What are the 4 L’s of a Student Portfolio?
The 4 L’s, defined by NACE for portfolio design, are Listen (observe a real user), Link (connect to a real job family), Learn (pick up a transferable tool or method), and Lead (own a decision). Strong portfolio pieces usually hit all 4.
5. Do recruiters actually look at student GitHub repos?
Yes, but only the first 60 seconds. A clean README, a passing CI badge, a “Built with” line, and a pinned “Highlights” issue are the four things that get you past 60 seconds. Prompt 25 is the audit list.
6. Should I list ChatGPT and GitHub Copilot in my “Built with” line?
In 2026, yes. NACE’s April 20, 2026 report found that 13.3% of entry-level jobs now require AI skills, and 28% of employers are seeking candidates who can use AI. Listing the tools is no longer a confession; it’s a signal.
7. How do I make a STAR story from a school project I didn’t lead?
Lead doesn’t mean you ran the team. “Lead” in the 4 L’s means you owned a decision. Pick the one decision you made - the chart type, the user research question, the deployment target - and STAR that. Use Prompt 4 to bank 3 STAR stories per project.
8. Is a personal website worth it in 2026?
Yes, but ship the 1-page version. Use Carrd, Framer, or Vercel with a GitHub Education domain credit. Prompt 26 is the brief. The bar is “exists, loads fast, has 3 projects.”
9. What’s the fastest way to find a real user for a student project?
Ask 5 classmates who aren’t in your major. They’re the only people in your life who will give you an honest “I don’t get it” in under 2 minutes. Pair that with Prompt 2 (Audience Reality Check) and you have a user by Friday.
A 30-day “3 portfolio pieces” sprint
If you have 30 days and want 3 finished portfolio pieces, here’s the schedule. Working backwards from launch day.
Week 0 (Days -7 to -1): PICK
- Day -7: Run Prompt 1 on your transcript.
- Day -5: Run Prompt 2 for the top 3 picks.
- Day -3: Run Prompt 3 on the weakest draft you have.
- Day -1: Finalize the 3 picks and the 1-sentence problem statement (Prompt 7) for each.
Week 1 (Days 1–7): SHAPE - Project A
- Day 1: Prompt 5 brief, Prompt 6 persona.
- Day 2: Prompt 8 stakeholder map, Prompt 9 trade-off memo.
- Day 3: Prompt 10 build plan, Prompt 11 stack.
- Day 4: Scaffold the repo. Use GitHub Education for the free pack.
- Day 5: First 2 days of Prompt 10 plan.
- Day 6: Mid-week check: does the project still pass Prompt 3?
- Day 7: Ship the rough demo (Prompt 14). It’s allowed to embarrass you.
Week 2 (Days 8–14): SHIP - Project A, then SHAPE Project B
- Day 8: Prompt 12 README + Prompt 13 review simulator.
- Day 9: Prompt 15 metric mining.
- Day 10: Prompt 16 outcome narrative + Prompt 17 postmortem.
- Day 11: Prompt 18 resume bullets + Prompt 19 case study skeleton.
- Day 12: Project B brief, persona, build plan.
- Day 13: Project B scaffold + first 2 days of build.
- Day 14: Mid-week Project B demo.
Week 3 (Days 15–21): SHIP - Project B, then SHAPE Project C
- Day 15: Project B README + Loom + metrics.
- Day 16: Project B resume bullets + case study draft.
- Day 17: Project C brief and build plan.
- Day 18: Project C scaffold + build.
- Day 19: Project C mid-week demo.
- Day 20: Project C README + Loom + metrics.
- Day 21: Project C case study draft.
Week 4 (Days 22–30): TELL - Site, posts, outreach
- Day 22: Prompt 25 repo polish on all 3 repos.
- Day 23: Prompt 26 personal site brief + 1-page build.
- Day 24: Prompt 27 LinkedIn launch post for Project A.
- Day 25: Prompt 24 “What I Wish I Knew” post for Project A.
- Day 26: Prompt 28 outreach DMs (send 5).
- Day 27: LinkedIn post for Project B.
- Day 28: LinkedIn post for Project C + site goes live.
- Day 29: Prompt 29 30-day update loop plan.
- Day 30: Run Prompt 4 STAR builder on all 3 projects and bank the stories.
Total artifact count after 30 days: 3 deployed projects, 3 case study pages, 3 LinkedIn posts, 3 Loom demos, 9 resume bullets, 9 STAR stories, 1 personal site. That’s a portfolio.
Common mistakes to avoid
Mistake 1: Picking 12 projects
Recruiters don’t read 12. They read 3, and only if the first 3 pass the 60-second test (Prompt 3). More projects ≠ more credibility. Pick the ones that score highest on the 4 L’s.
Mistake 2: Hiding AI use
In 2024, students hid ChatGPT use. In 2026, NACE’s data shows 13.3% of entry-level jobs require AI skills. Hide it, and you look out of date. Show it, and you look fluent. Add it to “Built with” (Prompt 12).
Mistake 3: Skipping the postmortem
The “What I’d do differently” section is the most underwritten and the most-read part of any case study. Use Prompt 17. Recruiters ask about it. It’s the cheapest interview prep you can do.
Mistake 4: Building for “anyone interested in”
Pick one person with a name and a job (Prompt 6). If your case study doesn’t open with a specific user, your project probably doesn’t either.
Mistake 5: Treating GitHub like a file dump
GitHub Education gives 5M students free tools. Use the README, the pinned issue, the description, the topics, the license, the CI badge (Prompt 25). The repo is the portfolio piece for engineers.
Mistake 6: Forgetting to tell anyone
The most polished portfolio that nobody sees is a 0. Run Prompt 27 (LinkedIn) and Prompt 28 (outreach DM) the day you launch, not the day you apply.
Mistake 7: Stopping after the first launch
NACE’s 2026 Spring Update shows the grad market is “fair” for 45% of employers. The bar is activity, not a one-time build. Use Prompt 29 to plan the 30-day loop.
Mistake 8: Designing for the course grade, not the user
Coursework is graded on rubric. Portfolio is judged on usefulness. Run Prompt 5 (Doubled-Diamond Brief) on every project. If you can’t fill the Discover/Define/Develop/Deliver squares, you’re not done.
Mistake 9: Writing the case study last
Write it first. The case study is the brief, the README, the resume bullets, the cover letter, and the LinkedIn post. If you can’t write a tight version, you don’t know what you built. Start with Prompt 16.
Mistake 10: Treating prompts as final output
ChatGPT prompts are scaffolding. The case study is yours. The metric is yours. The postmortem is yours. The “I learned” is yours. Run the prompt, throw away 40% of the output, and rewrite the rest in your own voice.
Final word
You don’t need 12 projects. You need 3, and you need 29 prompts that don’t waste your time. The Course-to-Project Pipeline is just PICK → SHAPE → SHIP → TELL, run with the discipline of a small studio and the honesty of a postmortem. The 4 L’s of a Student Portfolio (Listen, Link, Learn, Lead) are the rubric. NACE’s 2026 data is the scoreboard: 5.6% more grad hires, 13.3% AI-required roles, 70% skills-based hiring, 85% teamwork in the resume scan.
Use GitHub Education - 5M students can’t all be wrong. Use the Design Council’s Double Diamond for the brief. Use the prompts above for the build.
Now close the tab and go ship something. You’ll be a better candidate 30 days from now, and a much better one 90 days from now.