AI & Technical question

How does recommendation work in Instagram?

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Tests understanding of recommendation-system fundamentals: signals, ranking, and how quality gets measured in a feed context.

Clarify what's being recommended, Explore feed versus Reels, since signals differ.
Explain signal sources: explicit (likes, follows, saves) and implicit (watch time, scroll-past speed, comments).
Describe the two-stage structure: fast candidate generation narrows millions of posts, then a ranking model scores them.
Explain the likely ranking objective: predicted probability of meaningful engagement, not just clicks.
Discuss balancing diversity and freshness against pure relevance to avoid a stale, narrow feed.
Explain quality measurement: A/B tests on engagement and, importantly, longer-term retention.

Describes the two-stage candidate-generation-then-ranking architecture used by large-scale feed recommenders.
Distinguishes engagement metrics from retention, noting the risk of optimizing purely for clicks.
Mentions diversity and freshness as a deliberate counterweight to pure relevance.

Describing this as one single 'AI algorithm' with no structure.
Ignoring the risk of over-optimizing for short-term engagement at retention's expense.

Chapters of the AI PM course that teach what this question tests.