Modelling techniques

Recommendations and ranking interview questions

Recommendation questions are usually design questions. You might be asked to build a recommender for a streaming service with no data on brand-new users, and the interviewer wants to see how you break the problem into candidate generation, ranking and evaluation.

Along the way come collaborative filtering and matrix factorisation, content-based and hybrid approaches, ranking metrics such as NDCG and MRR, and the feedback loops that make popular items more popular. Good answers say how they’d measure success offline and then check it online.

Try one now

Tough

Design a recommender for a streaming service that has no data on brand-new users.

Free, no sign-up. The interviewer asks follow-ups on your reasoning, then rates your answer.

What the 7 questions cover

  • Ranking metrics (NDCG, MAP, MRR)3
  • Feedback loops and popularity bias3
  • Collaborative filtering and matrix factorisation2
  • Cold start2
  • Content-based and hybrid recommenders1
  • Candidate generation (two-tower retrieval)1
  • Learning to rank1

A question can cover more than one topic, so the counts add up to more than 7.

How the practice works

  1. The AI interviewer asks the question and follows up on your answer, the way an interviewer would.
  2. Stuck? Ask for a hint. Each one gives a little more away, so you can get unstuck without being handed the answer.
  3. Get a written debrief. Four ratings, each from Needs work to Interview-ready, tied to what you actually said, and what to study next.

More modelling techniques questions

See all categories