CVS Health Data Scientist candidates report SQL and Pandas screens, then statistics, machine learning and case-study rounds, with some contract processes starting with a recruiter or AI screen.
$130K
Avg. Base Comp
$173K
Avg. Total Comp
2-4 rounds
Typical Rounds
Not reported
Process Length
Candidates describing CVS Health Data Scientist interviews most often report SQL and Pandas screens first. Reported SQL topics include joins, CASE WHEN, subqueries, window functions, GROUP BY aggregation, and finding duplicates or NULLs with a LEFT JOIN. Difficulty ranges from easy LeetCode-style questions to medium-difficult, and one candidate said the Pandas work needed to mirror the SQL, with cumsum mentioned. One candidate described a prescriptions table where the task was total spend per patient, which suggests healthcare-flavored data is plausible.
After the coding screen, reports diverge. One candidate described a second technical round with CTEs, advanced joins, ML and stats questions, followed by a VP case study applying ML to a business use case. Another reported hypothesis testing, Type I and Type II errors, and ML theory, then a behavioral round. A third described an ML case study covering Logistic Regression, random forest and XGBoost tradeoffs, class imbalance and a full A/B testing walkthrough. Practice explaining your reasoning aloud, since candidates mention being asked to walk through their thought process and defend tradeoffs.
At least three candidates mention contract roles, and the first stage there may be an AI screen or a third-party recruiter call covering basic Python and SQL. Round counts reported range from two to four, so confirm the schedule with your recruiter. Experiences are mixed: some report offers, while others report a rushed SQL phone screen or definition-heavy questioning.
Synthesized from 16 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Cvs Health process.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Cvs Health
Write a query that returns all neighborhoods that have 0 users.
| Question | |
|---|---|
| Rain in N Days | |
| Button AB Test | |
| Flight Records | |
| Always Excited Users | |
| Bagging vs Boosting | |
| RMS Error | |
| Size of Joins | |
| Detecting ECG Tachycardia Runs | |
| Cumulative Reset | |
| Brain Cancer Treatment Outcomes | |
| Total Transactions | |
| Using R Squared | |
| Random Forest Explanation | |
| Causal Email Journey | |
| Valid Anagram | |
| Percentage of Revenue by Year | |
| Xgboost vs Random Forest | |
| Lasso vs Ridge | |
| Assumptions of Linear Regression | |
| Duplicate Rows | |
| Count Transactions | |
| Type I and II Errors | |
| Hidden Culprit | |
| Slow SQL Query | |
| Data Preparation for Imbalanced Data | |
| A/B Testing a Checkout Button Change | |
| Model Product Performance Degradation | |
| Overfit Avoidance | |
| SageMaker Deployment Architecture |
Synthesized from candidate reports. Individual experiences may vary.
Candidates report different openers: an HR conversation about profile alignment, an AI-assisted Python and SQL screen grounded in real-world context, a third-party recruiter call of about an hour, and an SHL assessment with Python, SQL, stats and probability. Expect the format to vary by role type, especially for contract positions.
Candidates report SQL covering joins, CASE WHEN, subqueries, window functions, GROUP BY and duplicate detection, paired with Pandas filtering, grouping, joins and cumsum. Levels reported range from easy to medium-difficult. Some were asked to show correct output or talk through their approach, and one phone screen reportedly allowed only about 12 minutes for three SQL questions.
Reports include Type I and Type II errors, hypothesis testing, regression, model evaluation, feature engineering, and comparisons of Logistic Regression, random forest and XGBoost. One candidate also described class imbalance and an A/B test walkthrough covering metrics, randomization unit, power analysis and scale-up decisions. Be ready for changed assumptions in follow-up questions.
Some candidates describe a case study where they were given a business problem and asked to propose a solution while interviewers checked their direction. One report said there is no single perfect answer, and a VP-level final round applied ML knowledge to a business use case. Prepare to explain tradeoffs and business impact clearly.
Several candidates report a behavioral stage, in some cases asking about previous experiences or project impact, and one described a final mix of behavioral, product thinking and technical discussion. Prepare a clear project walkthrough and examples of working with stakeholders. Outcomes reported include offers, rejections and candidates still waiting.
