Data Scientist, Causal Inference

United Kingdom
295 days ago
Data Scientist, Causal Inference

About the Role
At Deliveroo we have a world class data science organisation, with a mission to enable the highest quality human and machine decision making. We work throughout the company - in product, business and platform teams. We are uniquely placed to connect and calibrate quantitative decision making throughout Deliveroo, and have strong relationships with our business partners.

Our team members use technical skills from the whole spectrum of data science: building analytical tools; informing decision making at all levels of the business via bespoke and automated analysis; running experiments; performing causal analysis; informing planning and prioritisation with robust impact estimates; building production machine learning and optimisation models; and upskilling the entire company in data literacy and data driven decision making.

At Deliveroo we're always experimenting. We have hungry customers, eager riders, and busy restaurants in 12 countries and over 200 cities and giving them what they want and need involves continually testing new ideas. Whether we're working to improve our restaurant recommendations or looking to find a more efficient algorithm for routing our drivers, experimentation helps us make the right decisions for our users.

Experimenting at this scale presents some unique challenges and we're investing heavily in building a world-class platform for designing, deploying, and analysing product experiments. We're looking for experts in statistical inference and estimation to join our growing team of data scientists and help us develop innovative statistical solutions for industrial-scale experimentation.

Some of the problems we're working to solve:
  • How can we monitor possible interaction effects across our experiments? How can we account for such effects in the analysis of our experiments?
  • How can we improve our inference techniques to correctly account for the many statistical tests we calculate? Once we have chosen such a correction, how can we account for it at the power calculation stage?
  • How can statistical methods help us estimate the long-term impact of experiments?
  • How can we leverage modern statistical algorithms in order to identify any business-relevant heterogeneity of treatment effects?
  • How can we avoid running afoul the selection bias across the many experiments we run? Can we use shrinkage estimators to get closer to the actual impact of an experiment once rolled out?
  • How can we use the outcomes of previous experiments to improve our inference for a given experiment?

  • Investigating complex methodological problems and working collaboratively with a small team of experts to establish world-class standards for experimentation.
  • Leveraging your broad statistical awareness to proactively identify opportunities to add to and improve our existing experimentation methods
  • Prototyping and amending statistical methods to fit our specific circumstances
  • Prototyping the code and providing clear communication and written documentation for the engineering team

  • PhD in statistics, economics, econometrics, or a relevant field of applied mathematics
  • Broad statistical awareness, including familiarity with frequentist and Bayesian approaches, and a demonstrated ability for developing innovative experimentation and analysis methods
  • Familiarity with a scripting language, some proficiency with Python would be an asset

Data Science at Deliveroo

We are a small team, with very large impact, seeking to answer some of the most interesting questions out there. We move fast, we're always looking for new ideas and we're very transparent about the decisions we make and why we make them.

There are so many questions we need to answer and plenty more we haven't even encountered. How do data and technology help restaurants to grow as consumer habits change? How can we predict what someone wants to order for dinner long before the idea has even crossed their mind? At Deliveroo these are just some of the tough problems we are solving - and there is no challenge that cannot be yours. No solution is owned by a particular team, which means the scope for growth and personal impact is enormous.

Benefits and Diversity

At Deliveroo we know that people are the heart of the business and we prioritise their welfare. We offer a wide range of competitive benefits in areas including health, family, finance, community, convenience, growth, time away and relocation.

We believe a great workplace is one that represents the world we live in and how beautifully diverse it can be. That means we have no judgement when it comes to any one of the things that make you who you are - your gender, race, sexuality, religion or a secret aversion to coriander. All you need is a passion for (most) food and a desire to be part of one of the fastest growing startups in an incredibly exciting space.
Should We Build X New Product Or Feature?
Product execution framework.
Why Do You Want To Be A Product Manager?
Learn how can you become a product manager.
How To Deal With Clients
His course is gonna help you to manage your clients in a better way with some tips and tricks. Understand what kind of clients you have (or may find along the way) and behave accordingly! Learn to say no (and how) and use some persuasion techniques to interact with them.