Goldman Sachs Data Scientist interview typically runs 2 rounds: screening, technical/ML round. It usually takes about 2 rounds over a few weeks and is highly structured and technical.
$187K
Avg. Base Comp
$380K
Avg. Total Comp
2-3
Typical Rounds
2-4 weeks
Process Length
Our candidates consistently report that Goldman Sachs is far more interested in how you reason about machine learning than in whether you can recite a polished data science toolkit. Across multiple experiences, the same themes keep surfacing: loss functions, logistic regression, tree-based methods, and the tradeoffs between boosting and bagging. What stands out is the level of rigor around why a method fits a problem, not just what the method is called. One candidate specifically noted being pushed on the derivation behind a loss function, which is a strong signal that this team wants people who understand the math beneath the model.
A second pattern we’ve seen is that the interviewers seem to value clarity under pressure. Even when the questions shift into probability or statistics, the bar is not about speed alone; it’s about staying organized and explaining your logic cleanly. One candidate described a probability tree and Buffon’s needle problem, while another mentioned rapid-fire regression questions. That combination tells us Goldman Sachs is screening for candidates who can move between theory and applied reasoning without losing precision. The non-obvious make-or-break here is usually not a missing buzzword — it’s a small conceptual slip, especially when you’re asked to justify a modeling choice or explain an objective function from first principles.
Synthetized from 3 candidates 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.
The part that stood out most to me was the final round, which was two back-to-back 45 minute interviews. In the first one, I was asked probability questions that felt more like a math screening than a typical data science chat, including a probability tree and the Buffon’s needle problem. I had to talk through my reasoning carefully rather than just give a final answer, so it helped to stay organized and explain each step out loud. The second interview switched gears into coding plus rapid-fire statistics. That section moved quickly and covered regression concepts, so it felt less like solving one long problem and more like being tested on whether I could answer cleanly under pressure.
Earlier in the process, there was also a more standard screening style conversation that touched on background, interest in the role, and behavioral fit. The overall vibe was structured and fairly technical, with some emphasis on practical problem solving and some on how I think through uncertainty. I would describe the difficulty as moderate to hard, mostly because the probability questions were unusual and the statistics portion was fast paced. I didn’t feel like they were looking for memorized answers as much as clear logic and comfort with fundamentals. If you’re preparing for Goldman Sachs, I’d make sure you can explain probability trees and classic probability puzzles, and also be ready to answer regression questions quickly without needing to pause and think too long.
Prep tip from this candidate
Practice explaining probability trees and classic probability puzzles like Buffon’s needle out loud, since the final round leaned heavily on that style of reasoning. Also review regression concepts in a rapid-fire format so you can answer quickly under time pressure.
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 | |
| Bagging vs Boosting | |
| Rejection Reason | |
| Level Of Rain Water In 2D Terrain | |
| Google Maps Improvement | |
| Append Frequency | |
| Cyclic Detection | |
| Minimum Absolute Distance | |
| Comparing Search Engines | |
| Find the First Non-Repeating Character in a String | |
| Target Indices | |
| Production Model Monitoring | |
| How Many Friends | |
| SageMaker Deployment Architecture | |
| Impossibly Iterative Fibonacci | |
| Deciding Between Solutions | |
| Client Solution Pushback | |
| Company Acquisition Choice | |
| Alternative Vendor Tradeoff | |
| LRU Cache 1 | |
| Credit Score Estimation | |
| Using APIs for Downstream Tasks | |
| Direct Mail | |
| Risk Model for a Mortgage Bank | |
| Analyzing Multiple Data Sources | |
| 2nd Highest Salary | |
| Empty Neighborhoods | |
| Rolling Bank Transactions | |
| Employee Salaries |
Synthesized from candidate reports. Individual experiences may vary.
An early screening conversation focused on your background, interest in the Data Scientist role, and behavioral fit. Candidates described this as a standard introductory round before the more technical interviews.
The first technical round was heavily centered on machine learning fundamentals and theory. Questions covered loss functions, logistic regression, and tree-based methods such as boosting versus bagging, with an emphasis on explaining concepts clearly and justifying choices.
The final stage consisted of two back-to-back interviews. One focused on probability and math screening topics like probability trees and Buffon’s needle, while the other switched to coding plus rapid-fire statistics, including regression concepts.