Aleksas Kateiva
About Aleksas Kateiva
Aleksas Kateiva is an author who has significantly contributed to the implementation and optimization of recommender systems at Vinted, particularly using the Vespa search engine.
Aleksas Kateiva Title Author
Aleksas Kateiva holds the title of Author. His work includes co-authoring a blog post focused on the implementation and benefits of the Vespa search engine for Vinted, particularly in the realm of personalized second-hand fashion recommendations. This work highlights his contributions to the technical and practical applications of search engine technology in e-commerce settings.
Aleksas Kateiva Vinted Recommender System
Aleksas Kateiva contributed to the implementation of a 3-stage recommender system at Vinted, a leading platform for second-hand fashion. The recommender system leverages both explicit and implicit user preferences to tailor recommendations. He played a key role in optimizing the first stage of this system using approximate nearest neighbor search with embeddings, which improved both the relevance and efficiency of the recommendations.
Aleksas Kateiva Vespa Search Engine Adoption
Aleksas Kateiva was instrumental in the adoption of the Vespa search engine at Vinted. His work involved operational tasks and performance tuning, making Vespa a viable solution for item recommendation use cases. Additionally, he participated in benchmarking Vespa against Elasticsearch to evaluate vector search database capabilities. This benchmarking was crucial in validating Vespa's performance advantages.
Aleksas Kateiva A/B Testing Vespa Performance
Aleksas Kateiva played a significant role in setting up Vespa for an A/B test to assess its performance for Vinted's specific use cases. Through this A/B testing experiment, he explored the trade-offs between approximate and exact search methods, providing valuable insights into the optimal configurations for Vinted's recommender system.