Uber Analytics Engineer Interview Questions (2026)

1. 1Recruiter Screen30 minEasy

   Role targeting, level calibration and process expectations.

2. 2Technical Phone Screen60 minMedium

   One or two medium DSA problems (CodeSignal or live) with emphasis on correct, runnable code and edge cases.

3. 3Onsite Coding I60 minHard

   Practical problem such as building a rate limiter or an in-memory index, judged on working code and API cleanliness.

4. 4Onsite Coding II60 minHard

   Algorithmic problem often with a geospatial or streaming flavor, pushed to optimal complexity.

5. 5System Design60 minHard

   Design a real-time marketplace system (dispatch, ETA, surge) with hard follow-ups on scale, geo-sharding and failure modes.

6. 6Behavioral / Hiring Manager Round45 minMedium

   STAR stories mapped to Uber's cultural norms: ownership, bold bets, customer obsession and conflict handling.

1. Q1

   Treatment improves request-to-complete rate but worsens ETA and cancellation rate for Airport Trips. Walk through a launch recommendation

   HardStatistics & Experimentation RoundA/B TestingUber-specific

   Context: Make a decision under conflicting metrics and quantify tradeoffs for stakeholders.

   How to answer: A strong recommendation would involve acknowledging the trade-offs and proposing a phased rollout or a segmented launch. The candidate should suggest further investigation into the root causes of the worsened ETA and cancellation rates, potentially through qualitative analysis or deeper dive into trip characteristics. A key step would be to quantify the monetary impact of the improved request-to-complete rate versus the negative impact of increased cancellations and ETAs, possibly using a North Star metric like net revenue or driver utilization. Finally, the recommendation should include a clear decision framework for proceeding, such as A/B testing the treatment only on non-airport trips or implementing a dynamic pricing model to mitigate negative effects.

2. Q2

   Estimate the business impact of changing pricing, commission, delivery fee, or ad load for Uber Reserve. What assumptions and sensitivities would you model?

   MediumProduct Analytics & Business CaseBusiness CasesUber-specific

   Context: The interviewer is testing whether you connect metrics to profit, not just top-line growth.

   How to answer: A strong answer would first define Uber Reserve and its target user base (planned, premium trips). Then, it would outline a framework for estimating business impact, likely focusing on a revenue-minus-cost approach. Key assumptions would include price elasticity of demand, changes in driver supply/utilization, and operational cost impacts. Sensitivities would explore different elasticity values, market segment responses, and competitive reactions. The candidate should also mention potential qualitative impacts like brand perception and long-term user retention.

3. Q3

   Fraud, abuse, or policy gaming is suspected in Rides. Size the financial impact and propose an analytics approach to reduce it

   HardProduct Analytics & Business CaseBusiness CasesUber-specific

   Context: Balance loss prevention with false positives and user/partner experience.

   How to answer: A strong candidate would first clarify the specific type of fraud/abuse (e.g., driver-side, rider-side, promo abuse) and its suspected mechanism. They would then propose a multi-faceted approach to sizing the financial impact, including analyzing transaction data (rides, payments, promotions), identifying suspicious patterns (e.g., unusual ride durations, repeated cancellations, high promo usage), and potentially using control groups or A/B testing for policy changes. For an analytics approach, they would suggest developing anomaly detection models (e.g., supervised learning for known patterns, unsupervised for novel ones), building dashboards for real-time monitoring, and collaborating with product/operations to implement policy changes and measure their effectiveness through key metrics like fraud rate, revenue impact, and false positive rates. Finally, they would emphasize an iterative process of detection, intervention, and measurement.

4. Q4

   Evaluate the ROI of a loyalty, subscription, or membership benefit attached to Driver App. How do you avoid mistaking selection bias for program impact?

   HardProduct Analytics & Business CaseBusiness CasesUber-specific

   Context: Use cohorts, holdouts, propensity, causal design, and margin-based economics.

   How to answer: A strong answer would first define the specific loyalty program (e.g., Uber Pro, fuel discounts) and its target drivers. It would then outline a methodology for calculating ROI, focusing on incremental driver engagement, retention, and acquisition, offset by program costs. Crucially, the answer must detail strategies to mitigate selection bias, such as A/B testing with randomized control groups, difference-in-differences, or propensity score matching, to isolate the true causal impact of the program. Finally, it would discuss key metrics and potential confounding factors beyond direct program impact.

5. Q5

   You need to create a self-serve dashboard for Uber Eats that PMs and business teams will use weekly. How do you define metrics and prevent misuse?

   EasyDashboarding, Stakeholder & Hiring Manager RoundDashboardingUber-specific

   Context: Include metric glossary, data freshness, filters, caveats, and examples.

   How to answer: To define metrics, start by understanding the core business questions and user needs, then identify key performance indicators (KPIs) that directly address these. Clearly define each metric with its name, formula (e.g., 'Gross Bookings = Sum of (Order Total - Promotions)'), aggregation level, and relevant dimensions. To prevent misuse, implement a centralized data dictionary or glossary, establish clear data governance policies, and provide in-dashboard documentation or tooltips. Regularly audit dashboard usage and gather feedback to refine definitions and improve user understanding.

6. Q6

   Create an operational scorecard for Airport Trips using ETA and cancellation rate. Which leading and lagging indicators would you include?

   MediumProduct Analytics & Business CaseProduct AnalyticsUber-specific

   Context: Make it useful for daily operations, not just monthly reporting.

   How to answer: A strong answer would define an operational scorecard for Uber Airport Trips, focusing on key metrics like ETA and cancellation rate. It would propose specific leading indicators such as driver availability near airports, predicted demand, and surge multiplier, and lagging indicators like actual trip completion rate, driver earnings, and customer satisfaction scores for airport trips. The candidate should also suggest a structure for the scorecard, including targets, actuals, and trends for each metric, potentially segmented by airport or time of day.

7. Q7

   Given trips(order_id, user_id, created_at, status, gross_amount) and riders(user_id, signup_date), write SQL to build weekly first-completed ride cohorts and calculate retention in weeks 1, 2, 4, and 8 for Driver App

   EasySQL Technical ScreenSQLUber-specific

   Context: Hypothetical Uber analytics case for Driver App; schemas may be simplified for an interview. Use first successful transaction as the cohort date and exclude cancelled or test records.

   How to answer: The candidate should first identify the 'first completed ride' for each user by filtering `trips` for 'completed' status and ordering by `created_at`. Then, they need to determine the 'cohort week' for each user based on this first completed ride's `created_at` date, typically by truncating to the start of the week. Next, for each subsequent ride, they calculate the 'week number' relative to the user's cohort week. Finally, they will group by cohort week and calculate the count of distinct users who had a completed ride in weeks 1, 2, 4, and 8, dividing by the initial cohort size to get retention percentages.

8. Q8

   The trip_events stream contains duplicated rows from client retries. Write SQL to deduplicate events using event_id, user_id, event_name, event_ts, ingestion_ts, and then calculate request-to-complete rate for Uber for Business

   EasySQL Technical ScreenSQLUber-specific

   Context: Hypothetical Uber analytics case for Uber for Business; schemas may be simplified for an interview. Keep the latest ingestion_ts for duplicate event_id values and explain your tie-breaker.

   How to answer: The candidate should first address deduplication using a window function like `ROW_NUMBER()` or `QUALIFY` with `PARTITION BY event_id, user_id, event_name, event_ts` and `ORDER BY ingestion_ts DESC` to keep the latest ingested record. After deduplication, they need to identify 'request' and 'complete' events. Finally, they will calculate the request-to-complete rate by dividing the count of 'complete' events by the count of 'request' events, ensuring to handle potential division by zero.

9. Q9

   Using trips and drivers, write SQL to find transactions in Airport Trips that have no matching driver dimension record. How would you quantify the business impact of this data-quality issue?

   MediumSQL Technical ScreenSQLUber-specific

   Context: Hypothetical Uber analytics case for Airport Trips; schemas may be simplified for an interview. Dimension tables are slowly updated and late-arriving records may be valid for a short window.

   How to answer: The SQL solution involves a LEFT JOIN from `trips` to `drivers` on `driver_id`, filtering for `pickup_location = 'airport'` and `drivers.driver_id IS NULL`. This identifies airport trip transactions lacking a corresponding driver record. To quantify business impact, one would calculate the percentage of airport trips affected, the potential revenue loss from unbillable or unanalyzable trips, and the operational inefficiencies caused by manual reconciliation or delayed reporting for these trips. Further impact includes compromised data integrity for downstream analytics and machine learning models.

10. Q10

    Sessionize trip_events for Uber for Business: create sessions when a rider has a gap of more than 30 minutes between events. Return sessions per user, average session length, and events per session

    MediumSQL Technical ScreenSQLUber-specific

    Context: Hypothetical Uber analytics case for Uber for Business; schemas may be simplified for an interview. The recruiter is looking for a window-function approach rather than procedural logic.

    How to answer: The candidate should first order `trip_events` by `user_id` and `event_timestamp`. Then, calculate the time difference between consecutive events for each user using `LAG()` and `DATEDIFF()` or `TIMESTAMPDIFF()`. A new session starts when this difference exceeds 30 minutes. This can be identified by creating a session start flag and then using a cumulative sum over this flag to assign a `session_id` to each event. Finally, aggregate the results to calculate sessions per user, average session length (max_timestamp - min_timestamp per session), and events per session.

11. Q11

    Using daily metrics for Uber Reserve, write SQL to flag anomalies where request-to-complete rate is more than 3 standard deviations from a 28-day rolling average. How would you handle seasonality?

    HardAdvanced SQL & Data Modeling Deep DiveSQLUber-specific

    Context: Hypothetical Uber analytics case for Uber Reserve; schemas may be simplified for an interview. Interviewer is testing whether you can produce a first-pass anomaly table and know its limitations.

    How to answer: The solution requires calculating the daily request-to-complete rate (completes / requests) for Uber Reserve. Then, for each day, compute a 28-day rolling average and standard deviation of this rate using window functions. An anomaly is flagged if the daily rate is outside three standard deviations from the rolling average. To handle seasonality, one approach is to use a 7-day or 14-day lag for the rolling window, ensuring the window primarily includes data from the same day of the week, or to include a `WHERE` clause to filter for specific days of the week if the seasonality is very pronounced and consistent.

12. Q12

    A PM disagrees with your conclusion that a Driver App launch should not proceed. How do you handle the conversation?

    MediumDashboarding, Stakeholder & Hiring Manager RoundStakeholder ManagementUber-specific

    Context: Assess backbone, openness to new evidence, and ability to separate data from ego.

    How to answer: First, I would seek to deeply understand the PM's perspective, their underlying concerns, and the data or assumptions driving their disagreement. I would then clearly articulate my own conclusion, presenting the supporting data, methodology, and any identified risks or limitations. The conversation would focus on aligning on the objective criteria for launch readiness and collaboratively exploring alternative solutions or mitigation strategies. If a consensus isn't reached, I would propose escalating the decision with a clear summary of both viewpoints and their implications to a relevant senior stakeholder.

13. Q13

    Uber Reserve has a heavy-tailed distribution for net revenue per trip. Which summary statistics would you report and how would outliers affect the mean?

    MediumStatistics & Experimentation RoundStatisticsUber-specific

    Context: Mention median, percentiles, winsorization sensitivity, and business relevance of whales or enterprise accounts.

    How to answer: For a heavy-tailed distribution like Uber Reserve's net revenue, I would report the median, interquartile range (IQR), and potentially specific percentiles (e.g., 90th, 95th, 99th) to capture the spread and upper tail. The mean is highly sensitive to outliers in a heavy-tailed distribution. A few extremely high-value trips (outliers) would significantly inflate the mean, making it a misleading representation of the 'typical' net revenue per trip. This inflation could lead to overestimating average performance or misallocating resources.

14. Q14

    Uber surveyed active riders about Driver App and found high satisfaction. What sampling biases might make the estimate misleading?

    MediumStatistics & Experimentation RoundStatisticsUber-specific

    Context: Consider survivorship, non-response, channel, region, tenure, and incentive bias.

    How to answer: A strong candidate would identify selection bias, specifically self-selection bias, as a primary concern. Riders who are highly satisfied are more likely to respond to a survey, while dissatisfied riders might ignore it or have already stopped using the service. They should also consider non-response bias, where the characteristics of those who respond differ significantly from those who don't. Finally, survivorship bias could be at play if the survey only targets currently active riders, missing those who left due to dissatisfaction.

15. Q15

    Model Uber's trip analytics using facts and dimensions. What is the fact grain?

    MediumRound 2: Trip Lifecycle Data ModelingData Modeling

    How to answer: The fact grain should be one row per business-defined trip unless a lifecycle-event grain is explicitly needed. Dimensions such as city, vehicle product, and user attributes should attach through stable keys. The explanation should emphasize declaring grain first, because metric correctness and join safety depend on it.

Practice these with instant AI feedback in a live mock interview → Start a Uber Analytics Engineer mock

Strong DSA and scalable system design; prepare analytical/behavioral stories

Indicative Analytics Engineer pay in India: ~₹9–40 LPA (role-level range, not a Uber-specific figure).

Other roles at Uber

Analytics Engineer interviews at other companies