Professional AssociationsProfessional Associations: ACM SIGMOD
more informationMore information: LinkedIn | Google Scholar | DBLP | Curriculum Vitae
IBM-Research, Almaden, USA Jan 2016 − Present Research Staff Member, Manager: Fatma Ozcan
• Building next generation big data enrichment, management and cognitive analytics platforms for IBM Watson, focussing on ontology-based entity extraction, resolution, integration and querying of knowledge graphs.
University of Maryland, College Park, USA Aug 2010 − Dec 2015 Graduate Research Assistant, Advisor: Amol Deshpande
- Subgraph-centric large-scale graph analytics on Apache Spark. Designed and built NScaleSpark, a system based on a "think-like-a-subgraph" paradigm, where users specify computations to be executed against a large number of multi-hop neighborhoods or subgraphs of the data graph. The system has been built over the Apache Spark platform and uses a series of RDD transformations to extract and hold the relevant subgraphs in distributed memory with minimal footprint and enables efficient graph computations over large-scale graphs.
Neighborhood-Centric Graph Analytics in the Cloud. Designed and built NScale, an end-to-end system that supports the distributed execution of complex neighborhood-centric graph analytics over large-scale graphs in the cloud. NScale allows users to declaratively specify subgraphs-of-interest, which are extracted and instantiated in distributed memory, while minimizing resource consumption to make the system suitable for deployment in cloud computing environments.
Scalable Workload-aware Data Placement for Transactional Workloads. Addressed the problem of transparently scaling out transactional workloads on relational databases, to support database-as- a-service in a cloud computing environment. Designed and built an end-to-end workload-aware distributed data placement and transaction processing system for a large distributed database with an incremental repartitioning mechanism to deal with workload changes.
Microsoft Research, Redmond, WA, USA Jun 2014 - Aug 2014 Microsoft Research Intern, Mentors: Jonathan Goldstein and Badrish Chandramouli
• Scalable high performance in-memory incremental analytics on big data in the cloud. Designed and built a system on Windows Azure, for enabling temporal and atemporal analytical, interactive queries on large datasets in distributed memory. The system is built using a high performance in- memory streaming analytics engine called Trill. We provide a programmable API for querying and persisting distributed sharded in-memory datasets. The system incorporates a pluggable components design to provide deployment platform independence.
Microsoft Research, Redmond, WA, USA May 2013 - Aug 2013 Microsoft Research Intern, Mentors: Raghav Kaushik and Ravi Ramamurthy
• Database-as-a-service in the cloud. Worked on the problem of efficiently querying large-scale data in the cloud. More specifically, our goal was to enable and optimize the performance of complex analytical workloads over large-scale data using hardware support. Extended Microsoft’s SQL server to build a system based on a tightly coupled hardware-software co-design using a configurable hardware (FPGA) module, for providing an efficient database-as-a-service.
Microsoft Research, Redmond, WA, USA May 2012 - Nov 2012 Microsoft Research Intern, Mentor: Badrish Chandramouli
• Progressive big data analytics in the cloud. Designed and built a large-scale data-parallel progressive analytics system for Windows Azure, that uses streaming engines as progress-aware reducers to provide support for progressive SQL over big data in the cloud. The system models progress as a first class citizen and enables processing of progressive data samples to provide meaningful early results that are deterministic, with repeatable semantics and have explicit provenance.
North Carolina State University, NC, USA Jan 2010 - Jul 2010 Graduate Research Assistant, Advisors: Injong Rhee, Weny Wang
Alleviating data congestion in mobile distributed systems. Designed and developed an adaptive data flow mechanism using priority control for media arbitration to improve throughput in mobile wireless systems characterized by asymmetric uplink and downlink traffic under conditions of high load.
Performance analysis of applications implemented as linux kernel modules. Developed a transparent web cache as a Linux kernel module using the netfilter architecture and evaluated its performance against an http proxy server implemented at the application layer.
Indian Institute of Technology, Madras, India Aug 2005 - Jun 2006 Master's Thesis, Advisor: Timothy A Gonsalves
Medium access control for mobile distributed systems. Designed and implemented primitives for medium access control to provide a network access mechanism for a large wireless distributed system with long haul links between mobile routers forming an ad-hoc mesh network.