Google Data Scientist interview questions

125 Data Scientists have shared notes from their Google interviews, with the questions they were asked in each round. That is 6% of the 2,146 Google interview notes on Screna AI.

125 Data Scientist interview notes

Google·Data Scientist·Technical Phone Screen

Jul 2026
Google data scientist technical screen, one coding question focused on Monte Carlo simulation for hypothesis testing. Pretty niche for a DS role but made sense given the team.
  • Write code that simulates n coin flips many times, tracks the head counts across simulations, and uses that empirical distribution to compute a two-sided p-value for an observed number of heads. No closed-form stats formulas allowed.

“My first instinct was to reach for scipy.stats and they immediately said no.”

View Post

Google·Data Scientist·Technical Phone Screen

Jul 2026
Google Data Scientist technical screen, two fairly meaty coding problems back to back. The questions were more software-engineering-heavy than I expected for a DS role, lots of Unicode edge cases and streaming constraints that I hadn't really prepped for.
  • Write an anagram-checking function that handles Unicode normalization, ignores case/whitespace/punctuation, strips combining marks for English locales, and runs in O(n) time with O(alphabet size) extra memory without materializing the full normalized string.
  • Follow-up: how would you adapt the anagram checker to handle locale-specific rules like German 'ß' expanding to 'ss', and how do you handle right-to-left scripts without breaking normalization?
  • Implement a lazy, stable de-duplication generator that yields the first occurrence of each distinct element, supports a custom key function, and optionally treats all NaN floats as equal, in O(n) time and O(distinct keys) space.

“The streaming constraint is what tripped me up.”

View Post

Google·Data Scientist·Technical Phone Screen

Jul 2026
Google data scientist interview that went deep into a single stats/ML problem. The whole session was basically one extended question about a custom error metric, bootstrapping, and probability theory. More math-heavy than I expected for a DS role.
  • Given a CSV with country, actual revenue, and predicted revenue columns, implement a percentage RMSE metric defined as the square root of the mean squared relative error. Your implementation needs to handle zeros and negative values, support optional country-level weights, and be numerically stable.
  • Implement a nonparametric bootstrap with resample size equal to n to get a 95% confidence interval for the pRMSE. Justify why you use n as the resample size, and explain when you'd prefer a stratified or cluster bootstrap instead.
  • What is the exact probability that a single bootstrap resample, drawn with replacement from n observations, produces the exact same sample in the exact same order as the original? Give the formula in terms of n and explain why this probability is negligible in practice.
  • What are the limitations of using bootstrapping to estimate uncertainty for this percentage RMSE metric, specifically around heavy-tailed distributions and cross-country dependence? What mitigations would you consider?

“This tripped me up more than I wanted to admit.”

View Post

Google·Data Scientist·Onsite - Behavioral / Leadership

Jul 2026
Behavioral screen for a Data Scientist role at Google, focused on soft skills and working across teams. Pretty standard culture fit stuff, but the persuasion question had some teeth to it.
  • Tell me about a time you had to convince non-technical colleagues to act on a recommendation you made.
  • Walk me through a difficult situation at work and how you handled it.

“I structured it as situation, action, result and it felt okay but I leaned too hard on the 'what I did' part and glossed over the actual impact.”

View Post

Google·Data Scientist·Technical Phone Screen

Jul 2026
Google DS interview with a meaty product analytics case about a search metric drop. The question had about six layers to it and I kept feeling like I was one step behind the whole time.
  • Searching time per user per session dropped 35% over the past three months. A teammate wants to model T1 (time to first successful result) and T2 (time to giving up) as separate distributions. Critique that approach and design a rigorous analysis to find the root cause.

“This question is basically five questions stacked on top of each other.”

View Post

Google·Data Scientist·Technical Phone Screen

Jul 2026
Technical phone screen for a Data Scientist role at Google. One statistics question that felt deceptively simple but had a real gotcha buried in it.
  • Can bootstrap help reduce variance?

“I said yes immediately and started talking about how bootstrap gives you a variance estimate, which...”

View Post

Google·Data Scientist·Onsite - Cross-functional / Panel

Jul 2026
A pretty intense cross-functional scenario interview for a Data Scientist role at Google, centered almost entirely on Trust and Safety conflict management. Four meaty questions, all tied to the same case. Left feeling like I'd either nailed it or completely talked in circles.
  • You're an analyst on a Trust team and a senior Growth PM (with VP backing) wants to loosen an upload filter to boost daily active users before a launch. Your analysis shows a 25-40% increase in exposure to violating content and regulatory risk. Walk through your full conflict management plan: stakeholder mapping, aligning on goals, writing a one-page decision memo, and how you'd keep things civil when the room gets heated.
  • How would you design a reversible experiment or phased rollout that balances growth needs against safety risk? Include your unit of randomization, guardrails, stop conditions, and on-call escalation plan. What single metric would you use as a hard kill-switch and what's its threshold?
  • If you believe a directive is unsafe, how do you formally document your dissent, get an independent safety review, and escalate without blowing up your working relationships? What changes if the deadline is 48 hours out?
  • Give a real example of a time you managed a conflict across team boundaries. Cover who the stakeholders were, what was at stake, what you actually did, the measurable outcome, and what you'd do differently.

“This one took me a while to get my footing on.”

View Post

Google·Data Scientist·Onsite - Behavioral / Leadership

Jul 2026
Behavioral loop for a Data Scientist role at Google. Three big questions, all structured around ownership and impact. Felt more like a research defense than a typical interview.
  • Tell me about the most memorable project you've worked on. What was the problem, who did it affect, and what did you personally own from start to finish?
  • Describe the most challenging research problem you've tackled. What made it hard, how did you pick your approach, and what did you change about how you work because of it?
  • Walk me through a time you pursued a goal with real uncertainty around it. How did you break it down, stay on track, and handle things when they went sideways?

“This one felt open enough that I rambled for a bit before landing anywhere useful.”

View Post

Google·Data Scientist·Onsite - System Design / Architecture

Jul 2026
System design round at Google for a data scientist role, centered entirely on one meaty ML/retrieval problem. The question had a lot of moving parts and I felt like I was barely keeping up with the scope.
  • Given a company like Coca-Cola as a starting point, design a system that retrieves the top 20 most similar companies globally for sales prospecting purposes.
  • How would you handle negative sampling and blocking to make this system scale to 10 million companies, including cold-start scenarios for newly added companies?
  • Walk through how you would evaluate this system both offline and online, including how you'd measure sales conversion lift and guard against data leakage from subsidiaries or parent companies.

“This question is basically five questions duct-taped together.”

View Post

Google·Data Scientist·Technical Phone Screen

Jul 2026
Tough technical screen for a DS role at Google. Six questions, all ML heavy, and they really wanted you to go deep on the math and tradeoffs rather than just name-drop algorithms. Left feeling like I'd passed maybe four of the six cleanly.
  • Define binary logistic regression from scratch: write out the probability model, derive the log-loss and its gradients with respect to the weights and bias, and explain why the loss function is convex and what that means for optimization.
  • Compare L1 and L2 regularization in logistic regression across sparsity, multicollinearity, decision boundary geometry, and probability calibration. When does elastic net make more sense than using either one alone?
  • Under what conditions does logistic regression tend to outperform a random forest? Think about decision boundary linearity, high-dimensional sparse features, small sample sizes with many features, and situations where calibrated probabilities matter.
  • Your model is overfitting. Walk through concrete remedies specific to logistic regression, random forests, and gradient boosting separately. Also describe how you'd detect overfitting beyond just looking at accuracy.
  • Contrast random forests and gradient boosting across bias-variance tradeoffs, sensitivity to noisy features, hyperparameter sensitivity, handling of missing values natively, and training versus inference cost. Give a real-world scenario where each clearly beats the other.
  • Case study: you have 50k rows, 10k sparse binary features, 1% positive class rate, and significant temporal drift in the data. Design an end-to-end pipeline for both a logistic regression with elastic net and a tree-based model. Cover feature processing, regularization choices, time-based cross-validation, threshold selection, probability calibration, and how you'd fairly compare the two approaches. Call out the specific pitfalls you'd avoid.

“Felt okay on the model definition and gradient derivation but stumbled a bit explaining convexity in a way that felt satisfying.”

View Post