About

Sweta Singh is a highly experienced Senior Software Engineer with a specialization in database internals at IBM. With over 18 years of product development experience in Db2, Sweta has honed her skills in optimizing schedulers in databases and data processing engines like Spark. Her current focus is on multi-tenancy development, and she has worked on two incubation projects to improvise adaptive workload management in Db2 using Reinforcement Learning and create workload management in Spark. Sweta has a strong background in performance engineering, having spent the first 15 years of her career in this field. She has worked on performance engineering of several database features like workload management, integration of Spark with Db2, online reorg, JSON implementation, minimizing overhead of security features like database encryption, auditing, and Label Based Access Control, and Db2 monitoring. She has designed benchmarks from scratch and was a technical lead for the OLTP performance for Db2 releases. In this process, she often had to study core database services like Data Management services, Index Management, Buffer Management services, Runtime, and determine ways to optimize the codepath. Sweta has also worked extensively with customers on performance issues and PoCs, helping them with their storage design, schema design, and database/query tuning for their workloads. She has presented best practices for performance at multiple forums, including IOD and Insights conference in the US and IMTC/IDUG presentations in India. Working on challenging customer issues always opened doors for innovations, leading to her patents on query optimization and a method of quickly finding the root cause of performance issues using statistical techniques. Sweta is proficient in several technologies like Spark, Reinforcement Learning, Postgres, and ML. She is also experimenting with Machine Learning techniques to automate test case generation for complex C products like Db2. Using Abstract Syntax Tree (ASTs) of open-source C code repositories (Linux & Postgres), she determines if a vector representation of code (code2vec) can effectively represent complex C functions. Overall, Sweta's relevant experience of over 20 years in software engineering, research, and development makes her a valuable asset to any organization looking to optimize database performance and scalability.