All micro-internships
Micro-internship brief ≈10h in one sittingbuilder

Answer 5 business questions in SQL

SQL is still the most-demanded data skill in job listings. Load a real dataset into SQLite or DuckDB, write queries that answer 5 genuinely useful questions, and document where the AI’s SQL suggestions were wrong — because they will be.

Data AnalysisSQL & DatabasesPython for Data

Posted by The AI Internships

The work brief

  1. 01Pick a real dataset with at least 2 relatable tables (or split one CSV into two — e.g. orders + products). Load it into SQLite/DuckDB (Colab works).
  2. 02Write 5 business questions a manager would ask, at increasing difficulty: a filter, an aggregate, a GROUP BY, a JOIN, and one with a subquery or window.
  3. 03Write the SQL for each — AI-assisted is fine, but run everything and verify the results make sense against the raw data.
  4. 04Keep an honesty log: at least one case where AI-suggested SQL was wrong or misleading, and how you caught it.

What you’ll produce

4 deliverables

Submission standard

Submit the dataset description, a public notebook/repo with the working queries, all 5 questions + SQL + results, and the AI-correction log. Ensure the linked notebook shows all five queries running with their outputs, and optionally link the AI chat behind your correction log.

  • The dataset, its tables, and why these 5 questions matter

    Written responseRequired

  • Public notebook or repo where the queries run

    Public notebook/repo where all 5 queries run with their result outputs visible.

    Public linkRequired

  • All 5: question, the SQL, the result, one line of interpretation

    All 5: question, SQL, the actual result, and one line of interpretation — results must match the notebook outputs.

    Written responseRequired

  • Where AI’s SQL was wrong and how you caught it

    Written responseRequired

You’ll complete these inside your private workspace.

What you must submit as proof

This brief requires evidence an AI can’t fabricate.

  • Public share link to the AI chat where you got (and corrected) SQL suggestions

    Optional but rewarded: makes your 'where AI was wrong' log verifiable.

    Public linkOptional

Submissions without this evidence cannot be submitted.

Protect other people in your proof. Blur faces, names, phone numbers and email addresses before you upload, and refer to anyone you worked with by role or number ("Listener 1", "the stall owner"). Your proof is only ever used to check your work — it is never published, never appears on your certificate, and is never shown in your public portfolio.

How your work is evaluated

The passing benchmark is 70/100.

Working, escalating SQL

29%

The 5 queries run, are correct, and genuinely escalate in difficulty.

Question value

14%

The questions are ones a real manager would ask of this data.

Verification honesty

29%

Results were checked against raw data; the AI-correction is real and instructive.

Runnable queries + real correction

29%

The 5 queries must actually run in the linked notebook with visible outputs, and the AI-correction must be a real, specific catch (not a generic 'AI was wrong'). Non-running queries or a fabricated correction log are a fail.

How we grade your AI usage

30% 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.