Harshada Chavan Email & Phone Number

San Mateo, California, United States

Redwood City, California, United States

- Part of In-memory column store team- Bloom filters - Query acceleration using GPUs- AI Vector support @Oracle

Greater Minneapolis-St. Paul Area

Using GPUs for accelerating spatio-temporal query processingSpatial data computing is not only data intensive but also compute intensive. Existing systems addressthe data intensive part. However, not much attention has been given to the compute intensive nature of the problem. Meanwhile, advancements in Graphics Processing Units (GPUs) have established them as a promising platform to perform compute intensive tasks. However, moving computations from CPU to GPU involve non-trivial challenges in terms of algorithmic change and optimal hardware utilization.As part of my PhD research, I am currently developing a GPU accelerated, end-to-end system to address the compute intensive nature of spatial computations. Our system exploits enormous computing power of GPUs to parallelize the computations. This will be the first end-to-end spatialcomputing system accelerated by GPUs.The system is based on NVIDIA GPUs and is developed in C/C++ and CUDA

Madison, Wisconsin, United States

Worked on MS SQL Server:- Characterized TPCH queries with respect to varying degree of parallelism (DOP)- Auto-DOP generation for queries in a concurrent system

Redmond, Washington, United States

Worked on optimizing KD-tree search used in Bing search.

Mountain View, Ca

Develeoped MapReduce jobs for analytical applications.

As a database developer.Worked on a new release of product Sybase IQ (a columnar database)- Elan, that was focussed on making IQ a write optimized store.I worked in the query processing team (QP) and handled query optimization and query execution.- Designed and developed new algorithms for data access, selectivity optimization, metadata integration, predicate evaluation; to accommodate data streams from the new intermediate store.- As QP interacts with many other modules of Database engine, I got a good exposure of data indexes, data storage, versioning and metadata handling.- Worked on the performance improvement of Microqueries. - Language: C++

As a database developer.Worked on a new release of product Sybase IQ (a columnar database)- Volkl. This release added the support for User Defined Functions in IQ.- Designed and developed UDF Simulator. This tool is useful to customers for testing the UDFs before they are executed on a running IQ server. The challenging part of this project was to understand how query execution engine of IQ works and provide similar functionality in the simulator. - The tool was developed in C++.

Responsible for providing internet connectivity and network services like computational clusters, mail services and web applications to students and faculty. Also handled software installations and maintenance of computational lab.

Saarbrücken, Germany

Design and development of Transaction Manager for Unbundled Database Systems(under the guidance of: Prof. Gerhard Weikum) - Extended the Lock Manager of RDF3X (in C++) by implementing predicate locking.- Designed APIs for communication between Transaction component and Data component in unbundled

Phd Candidate

M.Tech (Cse)

Bachelor Of Engineering - Be