LinkedIn Senior Data Scientist Interview Experience — Phone Screen with Two Product Sense Questions and a Harmful-Content SQL

Two product sense questions.

Question 1: A climate summit is happening this month, and users' home feeds are full of climate-related posts. Some users have said they don't want to see these posts. How do you tell whether the platform should take action and adjust the home feed?

Question 2: An old classic. The home page is changed to show only posts from users you follow. What do you think of this idea, and how would you measure the impact?

SQL: one I had never practiced, about harmful content violations.

Part 1. The table looks like this:

member_id | violation_id | violation_date
1         | 1            | 2025-01-01
1         | 2            | 2025-03-01
2         | 3            | 2025-02-04
3         | 4            | 2025-05-28

If a user commits three violations within a 90-day window, we flag that user. How many users would this method flag?

Part 2. Same table, but with an added column for violation type (harassment, nudes, etc.). The flag definition changes: a user is flagged only if they commit three violations of the same type within 90 days. How many users would this method flag?