Uber Data Scientist Interview Guide

This interview guide has been written for people preparing for Uber’s data science hiring process. We break down what each round covers and what Uber interviewers are looking for. Our writing has been informed and reviewed by senior data scientists working at Uber.

TL;DR this is what Uber’s data science interview loop looks like:

Step 1: Recruiter screen

Step 2: Hiring manager screen

Step 3: Technical phone screen

Step 4: Onsite

• 4.1 SQL & Python
• 4.2 Product case
• 4.3 Experimentation
• 4.4 ML System design (for certain senior roles)
• 4.5 Behavioral

The SQL and behavioral rounds are pretty consistent across different teams, but the product and experimentation rounds will focus on the problems your specific team faces along with general marketplace issues. 

Uber recruiters tend to be quite transparent, so ask a lot of questions about the rounds you’ll face and what you’ll be tested on. They’ll ask you to walk them through your background and relevant experience.

What are recruiters looking for?

The recruiters want to know if your work experience and skillset match what the role calls for. They also want a clear answer of why you want to work at Uber. 

Questions you might be asked:

• Tell me about yourself.
• Why Uber?
• Why do you want to leave your current role?
• Are you interviewing with other companies? 
• If you have multiple offers, how are you going to choose which one to go with?

This will be a 30-45 min conversation and will focus on your past work and projects. Also expect one or two product-sense questions.

The question running through the HM’s mind is whether this person thinks like someone who belongs here. The HM will observe how you describe your projects and your product intuition.

Walk the HM through your past projects and focus on the business impact they had.

What might the Uber HM ask you?

• Tell me about a project where data changed a decision that would have gone differently otherwise.
• What do you know about the team you'd be joining?
• Walk me through the most technically challenging analysis you've done in the last year.

Before the onsite, you will have a technical screen on CoderPad or CodeSignal. Overall it will take around 60 mins, you have 30-45 mins to solve a couple of SQL problems and one Python/pandas question.

Post that, the interviewers will ask questions to test your statistical knowledge. The stats portion will test your basics and A/B testing depth. You could also be asked to design an experiment from scratch as well. 

These questions have been reported by Uber DS candidates for the tech screen:

• Given a ride_ratings table and a rides table, what are average ratings by day for users with 5+ rides, and is there a relationship between the number of rides and average rating?
• Find all Uber rides which happened on rainy days before noon. Is there a relation between rain and driver rating?
• Given two arrays of ride durations from a control and treatment group, can you figure out the difference in median with a 95% bootstrap confidence interval?
• How would you measure the impact of surge pricing on driver retention?
• What does a p-value of 0.03 mean? How would you explain it to a stakeholder without a statistical background?

The areas covered for this screening can be all over the place, be sure to ask for as much info as you can get from your recruiter, but keep in mind some candidates have received info that did not align with the screen.

For SQL be sure to practice on some time-series data using window functions and CTEs.

For the onsite there will be 4 to 6 back-to-back interviews, each one around forty-five to sixty minutes long. You should expect at least one round of each:

1. SQL & Python
2. Product case 
3. Experimentation
4. Behavioral

If you’re going for a senior role, you should expect an ML system design round and an extra behavioral round with an Uber leader.

You’ve already passed the first SQL screen, so expect harder problems and a higher bar here. The coding interview is also focused on your statistical understanding. 

Some candidates have been asked to implement a statistical test in code.

What kind of questions should you expect?

• Find every driver’s earnings per hour across different city-time combinations using window functions.
• Write a solution to find the cancellation rate of requests with unbanned users (both client and driver must not be banned) each day between "2026-06-01" and "2026-06-03" with at least one trip. Round the cancellation rate to two decimal points.
• Can you find the 7-day, 14-day, and 30-day retention by rider cohort signup date?
• Can you code the t-test from scratch in Python?
• Can you find suspicious driver behavior using trip patterns? 

Since this round mixes code and statistics, we recommend you refresh that muscle and write a t-test or bootstrap confidence interval in Python.

Interviewers want to see if you can take a complex question like “which drivers are more likely to quit?” and turn that into an optimized SQL query.

You’ll be paired with a PM for a 60 min case discussion that’ll focus on Uber’s problems and marketplace dynamics. 

Interviewers at Uber want to see how well you pick and design metrics. Also, if you account for all sides of the marketplace when proposing a solution, and if you can think through potential second-order effects on your own.

What questions do Uber interviewers ask in the product case round?

• At what price point should we set the cancellation fees? Can you evaluate what would be optimum for the marketplace - $1, $3, or $5?
• We’re deciding which new cities to bring Uber Eats into, what data would you need for this?
• You're launching a product in a new marketplace with no historical data. How do you solve demand prediction?
• Driver active hours dropped 20% in Barcelona. How do you investigate? 
• How would you measure the success of Uber Freight?

We also suggest you read a few pieces of Uber's engineering blog to understand how different teams at Uber approach business and technical problems.

A senior Uber data scientist who also coaches for us says you should always account for all sides of the marketplace:

For the experimentation round, expect a lot of questions on A/B testing, along with a big focus on real world experience on experimentation.

Uber interviewers want people who can think and design experiments based on the specific constraints Uber has to navigate. Concepts like CUPED, switchback experiments, spillover effects and multiple comparisons corrections also come up quite frequently.

What sort of questions do Uber interviewers ask?

• Do you think if we improve base driver pay by 10% it’ll improve supply in low-density areas? Can you design an experiment to find that out?
• The team was testing 20 different strategies to optimize pricing, only one of the results was statistically significant. What would you do?
• So, at Uber we’ve noticed network effects can mess up standard A/B testing for us, can you think of any reasons why? How would you design A/B tests? Or would you use any alternatives? 
• On our freight platform we connect cargo owners with truck drivers; how would you design an experiment to test if showing the ratings for cargo owners affects how fast drivers accept their jobs?

Prepare for this round in the context of Uber’s marketplace and how changes to one side can affect the other. For example, user-level randomization can fail for interventions like surge pricing or dispatch matching, so interviewers will want to see if you spot that.

If results are conflicting across the 3 sides of the marketplace the tiebreaker at Uber is money. As a senior Uber data scientist says:

If you do have this round, you’ll be expected to design an ML solution starting from the business problem.

Uber wants candidates who understand the business, and once given a business issue or metric can turn that into a concrete ML objective. 

How are you approaching the problem and breaking it down? What are you basing your decisions on? How do you choose features? How do you reason about data pipelines and how deeply do you think through the constraints? 

Uber DS candidates have reported these ML system design questions:

• Do you think you can predict if a driver is going to accept a job when they’re pinged? How would you build a model to do this? What features do you think are important?
• Can you design a ranking system that decides how many drivers to notify and which drivers to target? Do you think a simple radius-based approach works here?
• Let’s say we’re thinking of offering drivers cash advances against future earnings if they operate in certain supply-constricted regions. How would you set up the problem to find out at what advance amount a driver defaults with 5% chance? How would you translate that into a model?

A Bar Raiser might appear in this round, they usually influence the leveling. A weaker performance at the senior level could result in a down-level offer, and for lower level roles, in rare cases, a strong no from the Bar Raiser could even end in a no-hire decision.

Uber values are paramount for this round. Currently, Uber’s eight values are Go get it, Trip obsessed, Build with heart, Stand for safety, See the forest and the trees, One Uber, Great minds don't think alike, Do the right thing. 

Be prepared with stories that map to these values, and talk about them as Uber does. Uber’s values have shifted a lot since the company began, but one value that’s stayed constant, even though the name has changed, is a “Go get it” attitude. 

What are Uber Bar Raisers looking for?

Uber Bar Raisers usually ask behavioral questions but you might also be asked about technical stuff.

Uber interviewers want to see stories that demonstrate ownership, ambition, empathy, customer focus, and times where you focused on the little details as well as the overall strategy. Your ability to talk to and influence non-technical stakeholders will be considered strongly as well. 

Along with some advice for the bar raiser portion, this ex-Uber interviewer on Blind suggested steering clear of boilerplate questions and information. If you’re unsure of something, don’t try to talk your way through it and instead say what you know and what you don’t. Keep this advice in mind for all the rounds.

For the behavioral round, Uber DS candidates have reported these questions:

• Have you ever been asked to do something that wasn’t technically feasible? If a senior stakeholder pushes you to go through anyway, what do you do? 
• What would you do if two key metrics are moving in opposite directions?
• How would you handle a situation where a partner from another team disagrees with your approach?
• Have you ever had a major disagreement or conflict with a teammate or PM? Can you tell me how the situation was resolved?

Once the onsite loop is over, every Uber interviewer submits written feedback, and at the end, a hiring committee decides (a) whether to hire you or not; and (b) your level if they do.

Many candidates on Blind report being down-leveled so keep that in mind and confirm level expectations early on with your recruiter, and try to demonstrate ownership throughout the interview loop if possible. Some candidates have reported Uber will not agree to change levels after the loop if you don’t have another offer in hand.

Uber DS total comp ranges from $145K at DS I to $483K at Staff DS based on data from Levels.fyi. The base ranges from $125K to $217K. 

RSUs vest on a four-year schedule, they’re more front-loaded than at most companies. The most common schedule is 35% in year one, 30% in year two, 20% in year three, 15% in year four.