Weizhi Xu
Senior Software Engineer, Zilliz
Senior Software Engineer at Zilliz, with many years of experience in vector search. His work has been published in top international conferences and journals such as VLDB and SIGMOD. He won the first place in NeurIPS'21 Billion Scale Vector Search Competition. Currently, she is responsible for vector database indexing, filtering, and end-to-end performance related research and development, focusing on reducing system latency and increasing throughput while ensuring stability and accuracy. Vicky Hui holds a Bachelor's and Master's degree in Computer Science from the University of Melbourne.
Topic
Vector Database Milvus High Performance Practices
This talk will focus on the practice and optimisation of the indexing part of the vector database Milvus and its on-cloud product ZillizCloud. We will explore how to efficiently perform vector similarity search in large-scale data environments and its practical applications in traditional recommender systems and retrieval augmentation generation (RAG) under the wave of large language models. The presentation will provide an in-depth analysis of how to improve the query performance of Milvus by algorithmic and technical means to solve the challenges in large-scale, low-latency and low-cost vector search, and help the audience understand the core value of vector databases in AI and big data application scenarios. Outline: Introduction to vector databases Milvus and ZillizCloud and overview of application scenarios Bottlenecks in large-scale vector retrieval High performance implementation and optimisation of vector indexing Summary and Outlook
