Search Ranking Design
← Back to ML System Design Interview Problems
Design a search ranking system. Pipeline: query understanding → retrieval (BM25 + dense) → ranking (learning-to-rank: pointwise, pairwise, listwise) → re-ranking. Evaluation: NDCG, MRR, click-through rate.
Related
- Embedding-Based Retrieval (dense retrieval stage)
- Recommendation System Design (similar pipeline)