About
Jyoti Leeka is a Senior Scientist in SQL server currently focusing on query optimization.
I started working at Microsoft in 2017 as a post-doctoral researcher at Microsoft Research. My work at MSR focused on query optimization for distributed systems. During my postdoctoral tenure I was involved in two projects: Instalytics and Blitz. Instalytics enables efficient big data analytics for Cosmos by careful co-design of the storage layer with the compute layer. Blitz aims at augmenting SCOPE query optimizer (which was branched out of SQL Server some years ago) with rules and operators to produce the best plan. In late 2019 I joined GSL and got the wonderful opportunity to work closely with SQL Datawarehouse query optimizer team. With SQL DW query optimizer team I contributed to (i) bluemoon project which is a feedback infrastructure for DW, (ii) auto materialized views tuner, (iii) cardinality estimation feedback work, (iv) addition of query signatures, (v) query shape which provides graphical representation of distributed plan with its associated properties and helps improve time to resolution of ICM’s, (vi) query regression analysis tool that helps triangulate possible root causes of query performance regressions, (vii) workload analysis tool which helps classify and recognize workload patterns for better query execution times. I also worked with SCOPE team on Physical Design Recommendation which enables users to pick columns to sort on or partition on based on data consumption to avoid shuffle therefore improving processing time. I wrote two patents to secure IP for Microsoft as well as disseminated the ideas of this work in top-tier conferences like ICDE, VLDB, SIGMOD, SIGMOD Record, EDBT, CIDR, FAST, CIKM, etc. I also served as the demo chair of SOCC 2023 and program committee member of VLDB, ICDE, EDBT, CIKM, AI-ML SYSTEMS, CODS-COMAD, ACM Web Conference, VLDB Journal, Journal of Big Data, SOCC, etc. This year I am one of the proceeding chairs of SIGMOD 2024.
Research Area: Data management, Query optimization, Distributed systems, Machine learning for Databases, Workload Optimization in the Cloud.