Yue Wang

About Yue Wang

Vector Search, Data Integration, Databases, Big Data, Algorithms.

Experience

1. Principal Research Sde

   Microsoft

   Sep 2017 — Present · Redmond, WA, US

   Microsoft Research.• Led and developed Index Recommendation for Vector Search of AI Infrastructure using HNSW and DiskANN (Paper below) achieving 2× to 8× speed-up compared to baseline.• Designed and Delivered Index Recommendation of relation tables for Azure PostgreSQL (Link below), significantly outperforming other cloud competitors. Also prototyped Json index recommendation.• Led and shipped novel scale-out Fuzzy Deduplication in Dynamics 365 Customer Insights (Link below) using Spark and Scala on 200 million rows (Paper below). Proposed a distributed Agglomerative Hierarchical Clustering algorithm achieving 2× to 19× speed-up. Drove adoptions of scale-out Spark Fuzzy Join (Paper below) in Customer Insights (Link below) and Azure Data Factory (Link below)• Led and implemented in-memory Fuzzy Join in Power BI Desktop (Link below) and Power Query in Excel (Link below), and in-memory Fuzzy Deduplication in Table-Lint for Excel and Power Query Online (Link below), using Locality Sensitive Hashing. Proposed and developed novel SIMD-based Levenshtein edit distance algorithm as a building block, achieving 5× to 13× speed-up.• Prototyped an LLM-based dataset search system and evaluated various approaches, achieving 88% F1 score.• Deployed Add-Column-from-Examples in Power BI Desktop (Link below) and Power Query in Excel (Link below), which uses input-output pairs to learn data transformation programs.

Education

• University of Massachusetts, Amherst

  Ph.D.

• Fudan University

  B.S.

Skills

• Python
• Latex
• C#
• Java
• Software Engineering
• Delphi
• Computer Science
• Algorithms
• Distributed Systems
• Machine Learning
• C
• Programming
• C++
• Pascal
• Hadoop
• Databases