Assistant Software Engineer

Roshan R Sumbaly

Contact ********@*****.*** http://cs.stanford.edu/people/rsumbaly

Information +1-415-***-**** http://github.com/rsumbaly

Building large-scale systems and data-mining

Interests

Stanford University, USA 2008 - 2010

Education

M.S. in Computer Science

Specialization - Database Systems

Teaching Assistant - Information Retrieval & Web Search, Data Mining

BITS Pilani, India 2004 - 2008

B.E. (Honors) in Computer Science

Cumulative GPA - 10.0/10.0

Teaching Assistant - Computer Programming 1, Parallel Computing

LinkedIn, USA

Experience

Software Engineer (April 2010 - Present)

Working on Project Voldemort ( http://project-voldemort.com ) and Hadoop as a part of SNA

( http://sna-projects.com )

Yahoo! Inc, USA

Technical Yahoo! Intern (June 2009 - September 2009)

As a part of the Cloud Computing & Data Infrastructure team, incorporated various compres-

sion algorithms at three di erent tiers of the PNUTS / Sherpa distributed datastore, resulting

in decrease of average round-trip latency.

Stanford University, USA

Graduate Research Assistant (September 2008 - June 2009)

Worked in collaboration with the Computational Earth & Environmental Science group to

port various sparse complex matrix solvers to NVIDIA GPU Clusters using CUDA.

Hewlett Packard Labs, India

Research Intern (January - June 2008)

Proposed and built a prototype data integration middleware (based on Grid Monitoring Ar-

chitecture), for aggregation of HP s enterprise data. Data integration was achieved using RDF

& SPARQL.

Indian Institute of Science (IISc), India

Research Intern (May - July 2007)

Worked in the Grid Applications Research Lab on prediction of job queue wait time in batch

scheduled machines ( like Torque, LFS ) using historical logs. Proposed new metrics and

algorithms while also building a generic simulator for replaying logs to test new clustering

algorithms.

Bhabha Atomic Research Centre (BARC), India

Intern (May - July 2006)

Worked on a scheduling algorithm, based on back lling optimization and fairness policies, and

deployed it on a 512 node cluster in the Supercomputing Research Facility at BARC. Also

contributed to an inhouse distributed monitoring system.

International Conference on High Performance Computing (HiPC 2007)

AIGA - Arti cially Intelligent Grid Assistant

Developed a Grid based Question Answering system capable of mining answers from distributed

data-sets.

Published an article in IEEE Technical Committee on Scalable Computing (TCSC) Newsletter

titled Deployment of a Natural Language Processing system on a Grid

Stanford University, USA

Projects

Update Summarization

Built a system which generates a summary of a multi-document dataset based on the assump-

tion that the user has already read a given set of documents.

Opinion Mining over Large News Datasets

Developed metrics and algorithms to determine the opinion about people by mining New York

Times corpus ( 1.8 million articles spanning over 20 years )

Implemented using Aster Data s nCluster - a Map-Reduce based RDBMS with infrastructure

provided by Amazon EC2.

Supervised Machine Learning Classi ers for Usenet newsgroup messages

Implemented variants of classical classi ers like Naive Bayes, SVM, Decision Trees and Nearest

Neighbor methods. Analyzed various existing feature selection methodologies and proposed

new domain speci c features to enhance accuracy of the classi er.

Encrypted Tweets

Built a client side symmetric-key encryption system for Twitter using Greasemonkey.

Also build a proxy server capable of performing man-in-the-middle attack on SSL.

BITS Pilani, India

Analysis and Implementation of Load Balancing Algorithms in Distributed Environments

Simulation of variants of the classical balls into bins load balancing algorithms using SimGRID

Toolkit.

Personalization using Link Analysis

Implemented various link analysis algorithms on browsing history for personalized recommen-

dations.

Programming: C/C++, Java, Python, SQL, OpenMP and MPI Parallel Programming

Skills

Toolkits: Eclipse, Lucene, CUDA, Hadoop

Platforms: Linux, Windows, Solaris, Mac OS X

Worked on Amazon s EC2, S3, Elastic MapReduce and Google s App Engine. Managed AWS

resources ( $30K worth of computing time ) for 50 students as TA at Stanford

Recipient of CEES/RPSEA 2008-09 Fellowship for research in GPGPUs

Achievements &

Recipient of UC Berkeley Fellowship & Purdue University Graduate Fellowship for 2008-09

Awards

Recipient of Narotam Sekhsaria Scholarship for 2008-10

Awarded BITS Pilani Alumni Global 30 under 30 Award in 2009

Recipient of Dhirubhai Ambani Undergraduate Scholarship & BITS Merit Scholarship, for all

four years of undergraduate studies

Awarded the Gold Medal for highest GPA in 2004 batch of BITS Pilani

Led a team of four to win the National Runners-up Prize at Microsoft s Imagine Cup 2007 for

the project eduGRID

Founder & Student Coordinator, Linux User Group and CSD (Centre for Software Develop-

ment) at BITS

Coordinator, Conferences

Solaris and Open Solaris, Java : Now & Future, Web and Mobile Applications using Net-

Beans, University Days, Sun Microsystems

Microsoft Robotic Studio, Microsoft

Cluster & Grid Computing, CDAC : Centre for Development of Advanced Computing

LY.dvi

Contact this candidate