Data Analyst · Step 8 of 8 · 12 min
Using AI safely, and your portfolio project
You can use AI to speed up formulas and SQL without risking data, and you have a plan for a public project.
What AI is good for
Chatbots are useful for drafting formulas, SQL queries and Python snippets, explaining someone else’s query, suggesting chart types, and tightening a written summary. Many spreadsheet and BI tools now include AI helpers too.
Describe your table structure, not your data: column names, types, and a few made-up example rows. That is usually enough to get a working query.
Check everything against the data
AI can write a query that runs perfectly and answers the wrong question, for example counting rows instead of distinct customers, or using the wrong join. Test generated formulas and SQL on a small case where you know the right answer, and compare totals with a trusted report.
Never paste a number into a report because a chatbot said it. Numbers come from your data and your query, which you can rerun.
Protect the data
Do not paste customer records, personal details, salaries or confidential figures into consumer chatbots. Use the tools your company has approved, which usually come with data-protection terms, and follow its policy. When in doubt, anonymise or use fake sample rows.
Keep your work reproducible: save the final query or notebook, not just the chat. Someone should be able to rerun it without the AI.
Your portfolio project
Employers want to see you think. Pick a public dataset (Kaggle, government open-data portals, the BigQuery public datasets) and one clear question someone could act on.
Deliver four things: your SQL or notebook, a short cleaning log, a dashboard or a few honest charts, and a one-page write-up with the question, the answer, the evidence and the limits. Put it on GitHub, Tableau Public or a simple web page and link it from your CV. One finished project beats five half-done ones.
Try it
Plan your project with AI as interviewer
- 1.Choose a public dataset on a topic you care about.
- 2.Use the prompt below so the chatbot helps you plan, without seeing any private data.
- 3.Write the question, the deliverables and a deadline in a short note.
- 4.Book two hours this week to start.
I am a beginner data analyst building a portfolio project. The public dataset is [DATASET NAME AND LINK], with these columns: [COLUMN NAMES]. Ask me questions one at a time to help me choose one clear business question, then suggest a plan: cleaning steps, 3 SQL queries to write, 3 charts, and an outline for a one-page write-up. Do not invent results.