Goldman Sachs Data Scientist interviews, as candidates describe them, mix ML theory, probability puzzles, Python and SQL coding, and resume deep-dives. Reports differ in format, so prepare across all four areas.
$146K
Avg. Base Comp
$185K
Avg. Total Comp
2-4 rounds
Typical Rounds
Not reported
Process Length
Candidates describe Goldman Sachs Data Scientist interviews in different formats, so prepare broadly. One candidate reports two ML-focused rounds that were mostly theoretical, with no coding or case questions. Topics included loss functions, including derivations and which loss to use in a given setting, plus logistic regression and boosting versus bagging. That candidate cleared the first round and lost the second to a small mistake, so practice explaining fundamentals slowly and precisely.
Two other candidates describe a more quantitative path. One reports a final round of two back-to-back 45-minute interviews. The first covered probability, including a probability tree and Buffon's needle. The second combined coding with rapid-fire regression and statistics questions. Another reports a HackerRank assessment that mixed LeetCode-style problems with probability, statistics, and algebra. That candidate also describes a conditional-probability puzzle about a six-barrel gun and a superday of back-to-back panels, with topics shared in advance.
Another candidate reports four rounds: statistics and Python, then Python and SQL with LeetCode medium problems, then a hiring-manager conversation about resume projects, and finally a senior-manager culture round. In that account, interviewers probed why the candidate chose certain models and tools.
A practical plan is to rehearse loss-function derivations and ensemble tradeoffs aloud, work through classic probability puzzles, keep Python and SQL sharp, and be ready to defend every modeling choice on your resume.
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 Goldman Sachs process.
I applied for a Data Scientist role at Goldman Sachs and went through a multi-stage interview process that spanned several months. The timeline was surprisingly long—I waited about two months just to get a start date, which ended up working against me since I received another offer in the interim and decided to pursue that instead.
The process itself had multiple components. There was an initial online assessment through HackerRank that tested both coding and mathematical reasoning. I had to work through LeetCode-style data structures and algorithm problems alongside probability, statistics, and algebra questions. The combination felt unusual—not just pure coding, but a real blend of quantitative foundations and programming fundamentals. Then came the technical rounds, which included a coding session with an interviewer present where I had to write and discuss solutions in real time. The interviewers were generally clear and helpful, which made the experience less intimidating than I'd expected.
What stood out most were the probability and statistics questions. One that stuck with me was a puzzle about a six-barrel gun with two bullets loaded consecutively. The question was whether shuffling the barrel before the next shot would actually increase the probability of firing a bullet, given that the first shot didn't fire. It's the kind of problem that tests your intuition about conditional probability more than raw coding ability. Beyond that, there were general machine learning conceptual questions woven in, and the interviewers seemed interested in both my technical depth and my ability to think through problems methodically.
The superday format—multiple back-to-back interviews with different panels on the same day—was intense but well-organized. Topics were shared ahead of time, which helped with preparation. Despite the structured approach and supportive interviewers, I didn't receive an offer. In hindsight, I think the long hiring timeline worked against the process for me, but if you do interview here, be ready for that probability and statistics emphasis; it's not an afterthought in their technical evaluation.
Prep tip from this candidate
Focus heavily on probability and statistics puzzles—not just LeetCode algorithms. Goldman Sachs tests conditional probability and statistical reasoning deeply in their technical rounds, so practice problems like the Russian roulette variant and be comfortable explaining your probability reasoning in detail. Also prepare for the HackerRank OA format combining coding with quantitative problems, not just classic DSA.
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 Goldman Sachs
Given that it is raining today and that it rained yesterday, write a function to calculate the probability that it will rain on the nth day after today.
| Question | |
|---|---|
| Recurring Character | |
| Find the First Non-Repeating Character in a String | |
| Bagging vs Boosting | |
| Rejection Reason | |
| Size of Joins | |
| Google Maps Improvement | |
| Level Of Rain Water In 2D Terrain | |
| Append Frequency | |
| Minimum Absolute Distance | |
| Cyclic Detection | |
| Comparing Search Engines | |
| Target Indices | |
| Success Measurement | |
| Production Model Monitoring | |
| Portfolio Platform Architecture | |
| How Many Friends | |
| Messenger Service Design | |
| Impossibly Iterative Fibonacci | |
| SageMaker Deployment Architecture | |
| Shortest Path Algorithms | |
| Deciding Between Solutions | |
| The Longest Journey | |
| Client Solution Pushback | |
| Company Acquisition Choice | |
| Alternative Vendor Tradeoff | |
| LRU Cache 1 | |
| Credit Score Estimation | |
| Using APIs for Downstream Tasks | |
| Direct Mail |
Synthesized from candidate reports. Individual experiences may vary.
Candidates describe different openings. One reports a screening-style conversation on background, interest in the role, and behavioral fit. Another reports a HackerRank assessment mixing LeetCode-style data structures and algorithms with probability, statistics, and algebra. Your first stage may be either, so prepare for both.
One candidate reports two ML-based rounds that were mostly theoretical: loss function derivations, logistic regression, and boosting versus bagging. Another reports a statistics and Python round followed by a Python and SQL round with LeetCode medium problems. The mix may depend on the team, so prepare for both styles.
Two candidates report heavy probability content. One describes a probability tree, Buffon's needle, and then coding plus rapid-fire regression questions. Another describes a conditional-probability puzzle about shuffling a six-barrel gun. Candidates report that explaining reasoning step by step matters more than giving only a final answer.
One candidate reports a superday of back-to-back panels with topics shared ahead of time. Another reports a final round of two 45-minute interviews. A third reports a hiring-manager discussion of resume projects, then a senior-manager round on working style and ambiguity. Expect to defend model and tool choices and your fit with the team.
