Indexing Internals Interview Questions | Qdrant | NextRound

1What does HNSW stand for, and at a high level, how does it achieve sub-linear approximate nearest-neighbor search?Intermediate

2What do the HNSW parameters m and ef_construct control, and what trade-off do they represent?Advanced

3What does the query-time parameter ef (search breadth) control, and how would you use it to trade off recall against latency?Advanced

4Why might increasing m significantly improve recall on one dataset but barely help - or even hurt latency - on another?Advanced

5What problem does vector quantization solve, and what is the fundamental trade-off it introduces?Intermediate

6Compare scalar quantization, product quantization, and binary quantization in Qdrant in terms of compression ratio and accuracy impact.Advanced

7What are oversampling and rescoring in the context of binary quantization, and why are they necessary?Advanced

8What newer quantization options - beyond the original scalar, product, and binary trio - has Qdrant introduced to fine-tune the compression/accuracy curve?Advanced

9What is Inline Storage, and how does embedding quantized vectors directly into HNSW graph nodes improve disk-based search performance?Advanced

10What is a multivector point, and how does it differ from a point with several named vectors?Advanced

11How does late-interaction scoring (as used by ColBERT-style models) with MaxSim differ from comparing two single dense vectors?Advanced

12Why is late-interaction reranking typically applied to a small candidate set rather than the entire collection?Advanced

13Design a three-stage retrieval pipeline using dense retrieval, sparse retrieval, fusion, and ColBERT reranking. What does each stage contribute?Advanced

14Why does Qdrant set m: 0 on a named vector used purely for reranking (e.g. a ColBERT multivector)?Advanced