Interview essentials for data science

Interview loops test SQL, Python/pandas, statistics intuition, ML concepts, and product sense. This lesson consolidates themes from the full track for rehearsal.

Technical pillars

• SQL — JOINs, GROUP BY, window functions (ROW_NUMBER, LAG)
• Python — data structures, pandas groupby, clean functions
• Stats/ML — bias-variance, metrics, train/test, cross-validation
• Case studies — metric design, A/B test interpretation, funnel analysis

Behavioral structure

Use STAR (Situation, Task, Action, Result) for project stories: define business problem, your analysis choices, impact metric, and what you would do differently.

Whiteboard habits

1. Clarify input schema and row grain
2. State assumptions and leakage risks
3. Propose baseline then improvements
4. Discuss tradeoffs and monitoring

Topics to rehearse from this track

EDA workflow, missing data types, correlation vs causation, precision/recall, ethics/fairness, reproducibility, SQL-in-pipeline architecture.

Important interview questions and answers

1. Q: Leakage example in interview?
   A: Using post-click features to predict click—explain time-safe feature cutoff.
2. Q: Imbalanced classification metric?
   A: Discuss precision-recall or PR-AUC, not accuracy alone.

Self-check

1. Name four technical pillars for DS interviews.
2. What is STAR format?
3. Give one leakage example you can explain aloud.

Tip: Prepare one project story: question → EDA → baseline → metric.