How to choose a portfolio project
Choose a project you can finish in two to six weeks, that uses the stack in the jobs you want, and that has at least one hard part you can explain. A small finished project beats a large abandoned one every time.
Start from the job, not the idea
Read five job posts for the role you want. Note the languages, frameworks and practices that repeat. Your project should use two or three of them, not all of them.
Check it against four questions
- Can I get a working version in two weeks? If not, cut scope until you can.
- Is there one part that is genuinely hard, such as sync, auth, search or scheduling?
- Can someone use it from a link, without installing anything?
- Will I be able to explain every decision in an interview?
What to avoid
- Tutorial clones with only the colours changed.
- Projects that depend on a paid API you cannot afford to keep running.
- Ideas so big that the demo is a screenshot.
Make it finishable
Write the smallest version that is still useful and call it the MVP. Put everything else in a "later" list. Ship the MVP, then add one feature at a time.
Present it
- A README with a screenshot, a live link and the three hardest problems you solved.
- A short note on what you would change with more time.
- Tests for the logic that matters most.