
What three candidates report about the TransUnion Data Scientist interview: a recruiter screen, a conceptual hiring-manager conversation, and technical depth that ranged from missing values to marketing attribution.
$120K
Avg. Base Comp
$148K
Avg. Total Comp
3 rounds
Typical Rounds
Not reported
Process Length
Three candidates describe the TransUnion Data Scientist interview, and the most detailed reports point to applied modeling judgment more than coding trivia as the core of the technical conversation. One candidate describes a recruiter screen followed by a hiring-manager round of conceptual questions, with a panel of team members expected next. That candidate was asked how to treat missing values in an important feature of a credit card payment assessment, when to use ML versus logistic regression, how to select the best model, and which test to run to evaluate a prediction. Another candidate describes a one-hour hiring-manager screen covering data processing, modeling, SQL and Python, and says the interviewer interrupted often.
The third report is the outlier. This candidate says the questions went well beyond a standard applied ML screen into marketing mix modeling, multi-touch attribution, Bayesian statistics and time-series methods, with the statistical assumptions behind each model expected. That is one account, so check with your recruiter whether the team does marketing analytics work before spending much time there.
For preparation, practice walking through a modeling problem from data cleaning to model choice to evaluation, and say your reasoning out loud in short, complete chunks. Be ready to justify logistic regression against more complex models in a credit-style setting. Keep SQL and Python sharp as well, since one screen covered both. Two candidates mention communication friction or slow follow-up after the screen. Ask what each round will cover and what the next step is before you leave a call.
Synthesized from 5 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 Transunion 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 Transunion
How would you assess if a coin is fair after observing 8 tails in 10 flips?
| Question | |
|---|---|
| Assumptions of Linear Regression | |
| Missing Housing Data | |
| 85% vs 82% | |
| 1000 Sample Classifier | |
| Evaluate News | |
| Concurrent LLM Serving | |
| Previous NaN Values | |
| Finding the Maximum Number in a List | |
| Subscription Overlap | |
| Merge Sorted Lists | |
| Find the Missing Number | |
| Bank Fraud Model | |
| Prime to N | |
| Rectangle Overlap | |
| Hurdles In Data Projects | |
| Groups of Anagrams | |
| LRU Cache 1 | |
| Confidence Interval Explanation | |
| Google Maps Improvement | |
| String Subsequence | |
| Shortest Path Algorithms | |
| Singly Linked List | |
| Minimum Parking Spots | |
| Implementing the Fibonacci Sequence in Three Different Methods | |
| Nearest Common Ancestor | |
| Filling Supermarket Bag | |
| Decision Tree Evaluation | |
| Longest Increasing Subsequence | |
| Processing Large CSV |
Synthesized from candidate reports. Individual experiences may vary.
One candidate reports an initial recruiter screen that covered past experience and an introduction to the team and role. Treat it as a conversation about fit and logistics. It is also a good moment to ask what the technical rounds will cover and whether the role is remote or hybrid.
Candidates report a conversation with the hiring manager, in one case lasting about an hour. Reported topics include data processing, modeling, SQL, Python, logistic regression results, missing value handling, model selection, and the test used to evaluate a prediction. One candidate says the interviewer interrupted frequently, so practice giving concise, structured answers.
One candidate was told the next stage would be a panel with team members from across teams. Another was told a coordinator would set up a next round but never heard back, so it is unclear how consistently this stage happens. Separately, a third candidate reports harder questions on marketing mix modeling, multi-touch attribution, Bayesian statistics and time-series modeling, without saying which round they came in. Prepare the assumptions and tradeoffs behind those methods in case your team works on them.