Demo case file · System design
Social media company
Machine Learning Engineer
This is a demo case file: a realistic example written from commonly known interview formats, not a real candidate's interview experience.
Questions asked
- Design the system that ranks posts in the home feed for 50 million daily users.
- What would you use as training labels and features, and how would you evaluate the model offline and online?
- How do you handle brand-new users and brand-new posts that have no engagement history yet?
What to expect
Expected the full picture: candidate generation, then ranking, then serving within a latency budget. The interviewer pushed on the gap between offline metrics and real engagement, feature freshness, and differences between training and serving. Mentioning the risk of optimising purely for clicks earned a follow-up discussion.
Practice this exact interview.
Mockd plays the interviewer live, using these questions and notes, then grades your answers.
Start a practice session Browse all case files