Evolving Pinterest’s Embedding Retrieval Platform

Summary

Pinterest describes how its Manas distributed search platform is evolving embedding retrieval for a corpus growing toward tens of billions of embeddings. Traditional HNSW indexes require all data in RAM, while single-vector two-tower retrieval limits context-dependent relevance, motivating changes across quantization, SSD-based serving, and multi-embedding retrieval. Manas uses Scalar Quantization and Product Quantization across HNSW and IVF, with SQ preserving over 90% recall while reducing index size, and Linear Scaling plus SIMD cuts per-query computing resources by 10–15%. For disk-based serving, SPANN with PQ is reported as 4.5x faster than plain SPANN and over 3x DiskANN QPS, while a five-billion-embedding production evaluation saved over 40% CPU time with under 5% recall drop. The platform is also adding ColBERT-style late interaction using Sum of MaxSim, with a pilot in progress.

Takeaways

1. 1

   Scalar Quantization reduced HNSW index size by 59% and IVF index size by 75%, while recall remained consistently above 90% in the reported benchmarks.

2. 2

   SPANN combined with PQ was 4.5x faster than plain SPANN and achieved over 3x DiskANN’s QPS with one-third the latency, at a 5% recall drop.

3. 3

   Multi-embedding retrieval adds ColBERT-style Sum of MaxSim scoring, requiring Manas to parse multi-embedding queries and run multiple ANN searches simultaneously.