Gemini Interview Questions & Experiences

Gemini·Data Scientist·Technical Phone Screen

Jun 2026

Gemini data scientist interview with a pretty brutal quantitative case study on ACH fraud detection. Four connected sub-problems covering expected value math, A/B test design, confidence intervals, and sequential testing. The kind of thing where you either know your stats cold or you're just guessing at formulas.

Given a proposed ACH fraud detection rule with known recall and false positive rate, compute the monthly gross fraud loss without the rule, the loss remaining after the rule is applied, and the net savings once false-positive operational costs are factored in. Show all formulas and units.
Design a two-arm online A/B test to detect a 15% relative reduction in fraud loss per 1,000 credits at 5% significance (two-sided) and 80% power. State your distributional assumptions, derive the per-arm sample size, and justify your modeling choice.
A control arm observes 240,000 credits with a 0.12% fraud prevalence. Compute a 95% Wilson confidence interval for that prevalence and explain what it means for a go/no-go decision.
If you're reviewing A/B test results daily over 14 days, how do you control Type I error? Propose a sequential testing plan with stopping rules and explain how early stopping affects your primary metric.

“This part I actually felt okay about.”

View Post

Gemini·Data Scientist·Onsite - Behavioral / Leadership

May 2026

Brutal onsite question for a DS role at Gemini. One question, eight sub-parts, all about ACH fraud. Felt more like a staff eng or risk PM interview than anything data science adjacent.

Walk through a high-severity ACH fraud incident you led end-to-end: what was your first signal, what was the business impact in concrete numbers, what did you decide in the first 24 hours, how did you get Risk/Compliance/Legal/Eng/Support aligned when they were pulling in different directions, what controls did you ship and how did you roll them out, how did you think about false positive rate vs. dollar loss, what did you tell executives and regulators, what postmortem actions did you own, and what would you change in hindsight?

“Eight questions duct-taped into one.”

View Post

Gemini·Data Scientist·Technical Phone Screen

May 2026

Interviewed for a Data Scientist role at Gemini and got hit with a pretty gnarly algorithmic problem involving sliding windows and ACH credit events. The follow-ups pushed into streaming systems and correctness proofs, which felt more like a systems design round than anything data science-y.

Given a stream of ACH credit events with user IDs, device IDs, timestamps, and amounts, implement an algorithm that finds, for each user, the earliest sliding window of t minutes containing at least k events from distinct devices where each amount meets a minimum threshold. Return the window bounds and device IDs, or indicate no such window exists.
How would you redesign this as a streaming solution with bounded memory if events arrive partitioned by user?
Prove that your window advancement logic and distinct-device counting are correct.
What is the worst-case behavior of your solution for highly bursty users, and how would you mitigate it?

“This one wrecked me for the first few minutes.”

View Post

Gemini·Data Scientist·Onsite - Multi Round

Sep 2025

Gemini DS interview that was basically a full fraud modeling case study compressed into one session. They handed you a real ACH dataset scenario and expected you to walk through everything from labeling to experiment design to slide deck structure. Pretty intense scope for a single round.

How would you define fraud labels using ACH return data, and how do you handle look-ahead bias, late returns, and partial reversals within a 5-business-day label window?
Propose 3 to 5 interpretable fraud detection rules for ACH transactions, including exact definitions and thresholds.
How would you design an offline backtest for this fraud policy, including sampling strategy, time-split validation, leakage checks, and counterfactual estimation?
What primary and secondary metrics would you use to evaluate the fraud control policy, and what guardrail metrics would you track to avoid unintended harm?
How would you analyze heterogeneous treatment effects across subgroups like user tenure, country, and device cluster, and how would you control for false discovery?
Walk through your online experiment design for launching this fraud policy, covering randomization unit, power analysis, ramp schedule, spillover handling, and rollback criteria.
How would you build monitoring for this fraud policy and separate seasonal effects like end-of-month payroll cycles from actual treatment effects?
Outline the three slides you would present to stakeholders covering problem and baseline, proposed controls and risks, and expected value and ramp plan.

“This is where I spent too much time second-guessing myself.”

View Post

Gemini·Data Scientist·Technical Phone Screen

Sep 2025

Gemini data scientist technical round, three problems back to back covering window functions, device fingerprint ranking, and a Python union-find thing. The SQL parts were manageable but the Python question had a lot of moving pieces and I'm not sure I nailed the complexity analysis under pressure.

Write a SQL query using window functions and joins to find, for each user, the earliest rolling 24-hour window (ending no later than 2025-09-01 23:59:59) containing at least 3 ACH credit transactions where at least one of those credits has a return within 5 days. Output the window boundaries, count of ACH credits, count of returns within 5 days, and a net exposure figure defined as total credit amount minus total debit amount occurring between the window start and the earliest return timestamp.
Write a SQL query to rank each user's devices by how many returned ACH credits are linked to that device in the last 30 days, where a device is considered 'linked' to a credit transaction if it was the most recent login within 60 minutes before the transaction. Return only the top-ranked device per user with its fingerprint and returned credit count.
In Python, process a streaming login feed (events may arrive up to 5 minutes late and out of order) and cluster device fingerprints that differ by exactly one character using union-find. Then identify all clusters where the canonical fingerprint is used by 3 or more distinct users within any 7-day window ending on 2025-09-01. Return a list of tuples with the canonical fingerprint, window start and end, and distinct user count. Also state the time and space complexity and explain how you handle late events and ties in canonical fingerprint selection.

“This one took me a while to even parse.”

View Post