Customized Data Science Interview Challenges

Customized Data Science Interview Challenges

Not just modeling
Real data science mini-projects, ~4 hrs
Based on real-world-like simulated data

Not just modeling
Real data science mini-projects, ~4 hrs
Based on real-world-like simulated data

Test skills that actually matter

Design an A/B Test

Build Metrics

Product Sense

Machine Learning


Data Cleansing

Only hire data scientists after they proved they know how to provide business value

Challenge Samples

Product – A/B Testing

Should a company increase pricing based on the A/B test results?

Full challenge

Marketing – Emails

How to improve email open and click through rates?

Full challenge

Growth – Referrals

How to measure the impact of a new user referral program?

Full challenge

Get a step-by-step tutorial on how to build data science challenges. For Free.

This tutorial explains how to build a challenge that allows to objectively identify top candidates. It provides practical examples on how to define and evaluate certain skills as well as how to define criteria for passing the challenge.

Buy a personalized data challenge

Each challenge is unique and will never be shared with anyone else
You will also receive a sample solution to be able to quickly score candidates
You will receive challenge and solution one week after the purchase

Analytics challenge

$ 10000

A/B testing problem, involving:
Cleaning wrong data
Run a statistical test
Analyze output
Suggest next steps

Buy This

Core Data Science Challenge

$ 10000

Modeling open-ended task, involving:
Business Acumen
Data Cleansing
Machine Learning
Actionable insights

Buy This

30 day money back guaranteed


Email if you couldn’t find an answer to your question here.

Can’t I just use a project from a machine learning competition site?

Yes, you could, and actually should, if:

  • Your DS will not have to define the outcome variable
  • Your DS will not have to choose the right loss for the project (such as false positive vs false negative cost, for instance)
  • Your data has no bugs
  • Your DS will not be asked to come up with actionable insights 
  • Your DS will never design an A/B test
  • Your DS findings will not have any impact on product development
Who will build the challenges?

Giulio Palombo worked as a data scientist for several high-growth Silicon Valley tech companies, the last one being Airbnb.
In his career, he interviewed hundreds of DS candidates at top tech companies.
He also wrote the book the book A Collection of Data Science Take-home Challenges that has sold >5K copies in 18 months.
Here you can find all info about the take-home challenge book, testimonials, and more.

Where does data come from?

Data is simulated, perfectly resembling real data at top tech companies. It even includes bugs, put there on purpose to check candidate attention to detail.
Simulated data allows to create data exactly the way you want to test specific skills.

Challenge samples give an idea of the data format.

When will I receive the challenge?

One week after payment you will get the challenge as well as the solution template that allows to quickly score candidates.

How long it takes for the candidate to complete the challenge?

About 4 hours. That’s typically the optimal length. Challenges longer than that have very low completion rate. Shorter challenges make harder to objectively evaluate several skills, and therefore reduce the on-site to offer rate.

What percentage of candidates passing the challenge will receive an offer?

Per our experience, that will be slightly above 50%.

I need help with sourcing candidates or scoring them, can you help?

Sure. Can definitely help with that. Email for further info.

Can I choose the challenge data domain?

Sure. Just email with a description of the data domain you want for your challenge.