Table of Contents
Lyft, Inc. is a rapidly expanding ride-sharing company based in San Francisco, operating in 644 cities across the United States and 12 cities in Canada. Founded in 2012 as a part of a long-distance car-pooling business originally called Zimride, Lyft launched in Silicon Valley in 2013.
The company has quickly spread, expanding from 60 US cities in April 2014 to over 300 by January 2017. Now, Lyft has grown to over 23 million users, with a billion recorded rides as of 2020.
Lyft generates millions of data points daily and needs to scale out their data science and research science capabilities. Hence, a dedicated data science and business intelligence department is tasked with leveraging the most advanced analytics, machine learning, and big data (using AWS S3/AWS EC2) tools in providing business models and insights.
Interested in data science at Lyft's main competitor? Check out the "Uber Data Science Interview Questions" article on Interview Query!
The Data Science Role at Lyft
The data science capabilities at Lyft are split into three specific teams: Data Scientists, Research Scientists, and Machine Learning Engineers.
Data scientists are responsible for building analytics infrastructure, creating models, and setting up dashboards for self-service analytics. Originally branded as data analysts, the data scientist role at Lyft is much more focused on analytics and being embedded with product managers to drive product decisions forward.
Research scientists at Lyft function more as traditional data scientists and ship production code to work on the core machine learning projects, such as the estimated ride time and the pricing of each ride. Lyft research scientists work a lot on the automation engines that run the Lyft app and product.
Lastly, machine learning engineers at Lyft focus on building the infrastructure that is needed to host the complicated models that the research scientists build.
Data science roles at Lyft are tailored specifically to teams in different products and features. As such, the precise role and responsibilities will depend on the teams and products/features you are assigned to.
Whatever the teams you are working with, or product/features your team is assigned to, the role of a data scientist at Lyft will span across business analytics, modeling, machine learning, and deep learning implementation.
Lyft has a culture of creating an “open, inclusive, and diverse environment where members are recognized for what they bring to the table”. Lyft prefers to hire highly qualified applicants with 3 years plus experience in data analysis and visualization.
Other basic requirements for hiring at Lyft include:
- BA/BS in a quantitative field like statistics, economics, applied math, operations research or engineering. Advanced degrees are preferred.
- 3+ years of industry experience in a data science or analytics role.
- Excellent data analysis, modeling, Python, and SQL language (MySQL, PostgreSQL, SqlServer, Oracle) skills - able to write structured and efficient queries on large data sets.
- Proficiency in Tableau, Power BI, or similar data visualization software.
- Very comfortable building new data tables using ETL logic: building, managing, and fixing entire enterprise data models.
- Proficiency in workflow management tools such as Airflow.
- Exceptional communication skills with the ability to present findings & recommendations targeted to the audience in question.
What kind of data science role?
There is a dedicated data analytics and business intelligence department at Lyft, but depending on the teams and product you are assigned to, the job role and function may differ a little. Depending on the teams, the functions may include:
- Leveraging statistical modeling, machine learning, or data mining techniques to deliver actionable recommendations to CET leaders.
- Leveraging machine learning to automate tools to manage growth levers.
- Track and make reports on KPIs connected to central work streams through weekly, monthly & quarterly business reviews.
- Analyze the market-level impact of price changes across the marketplace.
- Work closely with product leads to build the most efficient tools, systems, and processes to manage pricing at scale
Data science teams at Lyft include:
- Autonomous Vehicle
- Dynamic pricing
- Growth team
- Rider experience
- Driver experience
- Fraud team
- Airport experience
The Lyft Interview Process
The hiring process at Lyft is similar to other big tech companies. The interview process starts with an initial phone screening with a hiring manager or HR, then a take-home challenge (with usually 24 hours delivery time) or technical screen. After successfully passing through both the take-home challenge and a 45 minutes long technical interview (two technical interviews in some cases), it is followed with five or six one-on-one interviews onsite, now virtually due to covid-19.
The initial screening is done via a phone call from an HR or a hiring manager. This interview is mainly exploratory and is resume-based. The main focus here is on assessing your background, especially past experience, roles, team dynamic, to determine if you are a potential fit.
The next step in the interview process is the technical interview phone screen with a data scientist. This interview lasts between 30 and 45 minutes, and the interview questions span around the fundamentals of probability, statistics, machine learning, business case study, definition of some operational KPIs, a walk-through of the maths from your hypothesis testing, and your technical/past project experiences.
The Take-Home Challenge
After completing the initial screening, you will receive a Lyft take-home challenge that you will have 24 hours to complete. Questions in the take-home challenge are case-study based questions (ridesharing dataset), and they comprise both technical and business side problems. In this challenge, questions typically span across different topics, such as churn rate measurement, optimization (using machine learning), designing/experimentation for recommendations, and creating a comprehensive report about your assumptions, limitations, and conclusions.
Note: A presentation of the take-home challenge will be done onsite.
After passing the technical screen, the next scheduled interview in the process is the on-site interview. This process comprises five or six one-on-one rounds of interviews with a data scientist or a team manager, each lasting for approximately 45 minutes. This is a half-day interview process involving whiteboard coding, project discussion with team managers and data scientists, business case studies, and statistical concepts.
In general, the cumulative interview process will look like:
- A presentation of the take-home data challenge– at this interview, the candidate is required to make a presentation of the take-home challenge submitted in an earlier part of the interview process.
Note: at this stage, you are expected to build a coherent story around your analysis while answering questions on what metrics you used and why you chose them.
- A business case study interview: questions in this interview are mainly open-ended, surrounding a real-life business case study. It is advisable to brush up on some of the unit economics metrics related to ride-sharing at Lyft.
- Statistics and probability with a data scientist: questions here revolve around hypothesis testing, such as the classic "coin got x heads during y flips”. It pays to familiarize yourself with business applications of key concepts, their variants and data manipulation using SQL.
- SQL/Python interview: this is a 45 minute long interview with a data scientist that involves whiteboard coding in SQL or R/Python and algorithm.
Brush up on your Python by reviewing the "Python Data Science Interview Questions" article on Interview Query!
- Core values/cultural fit interview with a product manager.
The aim of the interview process is to assess your experience with analytical concepts and design skill in providing business impact insights. Remember to brush up on your knowledge of statistics and probability (A/B testing), experimental design, and the business applications of key statistical concepts.
Also, reading up on key economic metrics and KPIs, algorithms, and models will be helpful. Practice lots of SQL and optimization problems on Interview Query , as these can better prepare you for the technical aspects of the interview process.
Lyft Data Science Interview Questions
- Describe how to engineer the heatmap telling drivers where to go. How would you define which areas will have high demand next and who do you want to go there?
- How do you model the impact of surge on demand and supply?
- Explain correlation and variance.
- Explain the best ways to achieve pool matching.
- How do you reduce churn on the supply side?
- What is the lifetime value of a driver?
- What are some of the different factors that could influence a rise in the average wait time for a driver?
- What optimization techniques are you familiar with and how do they work on a basic level? How would you find the optimal price given a linear demand function? Take a derivative of a quadratic function.
- How do you draw a uniform random sample from a circle in polar coordinates?
- Find expectations of a random variable with a basic distribution. How would you construct a confidence interval?
- How would you estimate the probability of a user ordering a ride? What assumptions do you need in order to estimate this probability?