All micro-internships
Micro-internship brief ≈14h across 2 daysbuilder

Add another language to something you shipped

Many users are excluded by English-only apps — localization is a real engineering skill with real judgement calls. Take a project you already shipped and add a working second-language toggle for a language you or a reviewer speaks: AI drafts the translation, you make it sound native.

AI Application DevelopmentGraphic DesignProduct Localisation

Posted by The AI Internships

The work brief

  1. 01Pick one of your shipped projects (landing page, game, tracker) and a language you actually speak.
  2. 02Engineer it properly: extract all user-facing strings into a translations structure, add a visible toggle, persist the choice.
  3. 03Translate with AI as a first draft, then fix it with a fluent speaker: tone, forms of address, what stays in English (buttons? technical terms?), and text that overflows the layout.
  4. 04Deploy the bilingual version and test it with one native-speaker who isn’t you.
What you’ll produce5 deliverables

Submission standard

Submit the live bilingual URL, the repo, a side-by-side screenshot of both languages, and your translation decisions log. Name the real native speaker (not you) who reviewed the translation and one thing they corrected; your repo must build on your earlier project's history.

  • Live URL with the working language toggle

    Public linkRequired

  • Public repo showing the string extraction and toggle code

    Public repo — commit history must show the string-extraction and toggle added on TOP of your earlier project (dated commits building on prior history), not a single fresh dump. A one-commit repo fails the authenticity gate.

    Public linkRequired

  • The same screen in both languages (one screenshot per language)

    Image uploadRequired

  • Your decisions log: what AI translated wrong, tone choices, what you kept in English and why, layout breakages

    Written responseRequired

  • Who reviewed it (their relationship to you and how they know the language) and the one thing they corrected

    Describe them by relationship, not name — "my aunt, who has spoken Marathi all her life". The specific fix is what counts; generic "a friend checked it" fails.

    Short answerRequired

You’ll complete these inside your private workspace.

How your work is evaluatedPass mark: 70/100

The passing benchmark is 70/100. Final score: Output 50% · AI process 30% · Evidence 20%. Required proof checks must also pass before certification.

Output criteria

Working toggle

40%

The live site switches languages fully and persists the choice.

Translation quality

40%

The translation reads native — fixed by a fluent speaker, not machine-dumped.

Localization engineering

20%

Strings properly extracted, not find-and-replaced; layout issues handled.

Required proof checks

Real i18n on a real prior build

Fails if the live toggle is broken/dead, the repo is a single fresh dump with no prior project history, or the native-reviewer detail is generic or absent. A relationship plus a specific correction is what is required — a reviewer named only by relationship is fully compliant.

Must pass
How we grade your AI usage30% of your score

Using AI is the point — it’s the skill this certificate proves. You’ll answer three short questions about how you used it: what you asked, what was wrong with its first answer, and what you changed. Specific, honest answers score high. “I pasted the brief and submitted the answer” scores near zero.