Competency rubric
| Area | Strong evidence | Weight |
|---|---|---|
| SQL/data | Correct grain, joins, windows, quality checks and reproducibility | 25% |
| Analysis | Sound measures, uncertainty, alternatives and limitations | 20% |
| Semantic model/BI | Relationships, measures, UX and access control | 20% |
| AI validation | Checks generated queries/claims and records corrections | 15% |
| Communication | Concise narrative linked to a decision | 20% |
Representative questions
A dashboard total differs from finance. What do you check first?
Clarify definitions and period, then inspect grain, joins, filters, late data, currency/tax treatment and measure logic before changing the visual.
When is average a misleading summary?
Discuss skew, outliers, subgroup differences, denominators and the decision the statistic supports.
How would you validate an AI-generated SQL query?
Review schema and grain, execute on controlled cases, compare totals, inspect null/duplicate behavior, test filters and record corrections.
What makes a semantic model AI-ready?
Clear names/descriptions, trusted measures, explicit relationships, security, representative queries and tested ambiguity.
Take-home assignment
Analyze a fictional customer-support dataset and recommend one operational change. Submit SQL, quality notes, a semantic model, two-page dashboard and a one-page memo. If AI is used, include a validation log.
| Dimension | Weak | Strong |
|---|---|---|
| Question | Explores without a decision | Defines user, measure and decision |
| Data | Assumes source is clean | Documents grain, quality and limitations |
| Analysis | Reports correlations as conclusions | Tests alternatives and uncertainty |
| Dashboard | Many charts, little hierarchy | Focused story with reusable measures |
| Recommendation | Generic | Specific, evidence-linked and measurable |
Fourteen-day plan
- Days 1–3: SQL joins, CTEs, windows and debugging.
- Days 4–5: descriptive statistics and uncertainty.
- Days 6–8: star schemas, DAX measures and RLS.
- Days 9–10: dashboard critique and business storytelling.
- Days 11–12: AI-assisted analysis with validation logs.
- Day 13: timed take-home simulation.
- Day 14: presentation and behavioral practice.
Questions are representative and are not attributed to a particular employer. Last reviewed 12 August 2026.