Vinted Adopts Vespa For Homepage Recommendations
Vinted is adopting the Vespa search engine to serve personalized homepage listing recommendations after evaluating Faiss and Elasticsearch in summer 2022. Benchmarks on ~1M documents showed Vespa delivered 3.8x higher indexing throughput and handled 8x more RPS with a P99 of 26ms versus Elasticsearch's 110ms at 250 RPS. Vinted deployed Vespa in an AB test on three 56-core servers to improve filtered retrieval and operations.
Key Points
- 1Replaced Faiss with Vespa to enable ANN retrieval with metadata prefiltering and real-time updates
- 2Benchmarks showed Vespa offered 3.8x indexing and 8x higher RPS with lower P99 latency
- 3Deploying Vespa enables reliable filtered recommendations under user constraints and simplifies operations
Scoring Rationale
Strong empirical benchmarks and practical deployment details justify impact; limited originality beyond engineering trade-offs and single-company evaluation.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
