Sumit Kumar
Bellevue WA 98005 USA
Phone +1-425-XXX-XXXX
Email [@]



  1. Integrated Masters in Maths & Computing, Indian Institute of Technology, Kharagpur(2000-05)

    Thesis title: An Intelligent Emergency Patient Monitoring System using distributed paradigms
    Advisor: Prof. A. K. Majumdar and Prof. J. Mukhopadhyay, Dept of Computer Sc. & Engg., IIT Kharagpur
    Summary: Implements an emergency patient monitoring system for heterogeneous set of devices (mobile phones, PocketPC etc) with intelligent notification mechanism. My architecture, based on mobile agents and tuplespaces, is able to provide unified content based notification service at the users' end, irrespective of their device type and corresponding device-specific limitations (storage, environment, UI, connection and computing capacity). The architecture is generic in its nature and can be used as a middleware to implement device-independent collaborative on-demand services. The Java based implementation provides platform independence and multiple choices for devices.


Work Experience

  1. Sr SDE at Amazon Aurora, Seattle (Jul 2016-Till date)

    Created automation for patching OS libraries on tens of thousands of hosts in the storage fleet without effecting service uptime.
    Implemented handling for s3 LSE without degrading db performance.
  2. Sr SDE at AWS Elastic MapReduce, Seattle (Oct 2013-Jul 2016)

    Simplified multiple internal EMR patches in Hadoop 2.4.0, Hive 13 and contributed back to open source. Imported several patches from open source into EMR's offerings. Improved processing performance for Hadoop and Hive by optimizing listing of input data on S3 and rename and delete of intermediate outputs. Improved performance of "msck repair table" for EMR Hive 13 on S3 by 10x of standard Apache implementation, this was even 2x better in performance of EMR's private feature "alter table recover partitions". Made several internal improvements on Hive 13 release process for EMR.
    Implemented EMR Kinesis connector (Hadoop, Hive, Pig and Cascading) and user facing samples to get started. Helped on-board several customers onto EMR using this connector. Implemented open source Kinesis Log4J Appender.
    Implemented several internal improvements for EMR as a platform. Implemented compression of logs that reduced the cost for logs and debugging by 10x.
  3. PMTS at iLabs(R&D), [24]7 Inc (formerly 24/7 Customer), Bangalore (Jul 2012-Sep 2013)
    Dev Manager at iLabs(R&D), [24]7 Inc (formerly 24/7 Customer), Bangalore (Sep 2011-July 2012)
    Tech Lead at iLabs(R&D), [24]7 Inc (formerly 24/7 Customer), Bangalore (July 2009-Sep 2011)

    Implemented log collection agent in scala that watches for application logs, transforms them incrementally and pushes them to a Kafka and Zookeeper based log collection platform. Implemented a data export application to fetch processed data from these platforms to consuming application databases in a scalable manner. Worked on Hadoop, Hive, Oozie and MySQL customization for batch data and real-time platform setup.
    Led a team of 3 to work on the front end features for auto completion, drupal based customizations to provide facebook like workspace and backend services to support these features. Designed dojo based object listener framework in javascript for the chat console in PxAssist.
    Led a team of 3-4 developers to develop Hive based reporting framework for PxOE and WoW!Px. Amazon EMR was used for generating these reports onto Amazon S3 and then a separate software uses this data to dump into the DB for reports. Used dynamic partitioning based approach to generate daily and weekly frequency reports. We solved sessionization of user data spread over 2 weeks and longer.
    Designed and developed an engine for visitor tracking and behavior evaluation (PxOE) on websites. Implemented a finite state machine based rules engine with parts of the same on client side and intelligence on the server side. By design all the IP was protected on the server side. Implemented client side framework for user data collection, finite state machine and rules evaluation. Implemented core features for browser side storage, cross domain communication, widget design and flow orchestration based on HTML5.
    Designed and developed a presentation and utility framework (WoW!Px) in native javascript for creating and deploying generic widgets on top of PxOE. PxOE and WoW!Px were deployed on 5 leading brands of US and Australia and gave a jump of 20% revenue for them within a month of deployment. A patent is pending for PxOE and WoW!Px idea and implementation.
    Designed a browser based chat console and implemented parts of XMPP based chat server backend. Designed and implemented the front end to support client specific feature customizations and license based feature enablement with a multi-tenant approach on a single deployment. Implemented server side REST framework. Implemented Cometd and Bayeux protocol based communication framework between the front end and the backend. Participated in security and scalability design.
  4. Senior S/W Engineer, Y! India R&D, Bangalore (July 2008-July 2009)

    Worked in Audience Global Platform group on Hadoop based platforms. Contributed to design and development of an open source apache project in Hadoop-Common codenamed Oozie - a workflow engine for managing and coordinating jobs running on Hadoop, including HDFS, Pig and MapReduce. Also worked on data acquisition and syndication from Reuters for Y! News end-to-end on hadoop based internal platforms codenamed Harvester and Pacman.
  5. Senior Member Technical Staff, Oracle India Server Technologies Division, Bangalore (May 2006-July 2008)

    Worked in Oracle Web Conferencing team that designed and developed streaming and transcoding solutions for sharing a presenter's desktop and voice in "view only" mode to standard players (Windows Media on Windows, QuickTime on MacOS and Flash on browsers). Developed a j2ee based streaming server that implements RTSP (RFC 2326), SDP (RFC 2327) and RTP (RFC 3550-1) and can stream data from any format to Windows Media Player and QuickTime Player using their respective proprietary extensions of the standard protocols. Also developed a java library that implements ASF and FLV specifications and can transcode data from any format to ASF and FLV formats. Worked on full development cycle of the product including research (byte level reverse engineering of then not public specifications), design and development from scratch, sanity testing and performance improvement.
  6. Associate S/W Engineer, Computer Associates, Hyderabad (May 2005-May 2006)

    Working in a newly formed "Platform Technologies Group" within the company. The group is responsible for porting existing CA applications to Unix (Solaris/HPUX/AIX) platforms. My contributions to CA within this small duration of 1 year, has been recognized resulting in a midyear performance award and a year end appreciation award (first in the history of CA-ITC). I along with my team have been able to deliver six different alpha and beta deliverables beating the deadlines. The final deliverable consists of 35+ individual product ports (that we did) with around 10 open source ported products, summing up to a 700+MB installable consumer product for each platform.
  7. Student Associate, TeleMedicine Lab, IIT Kharagpur (May-Jul 2003)

    Implemented TeleMedicine software for PocketPC. Developed using eVB and MSSQL Server 7.0, it provides an easy-to-use interface for EMR-browsing, EMR-analysis and teleconsultation over a TCP/IP connection and supports both online/offline operations. A paper was presented at Asia Pacific Telecommunity (APT) Telemedicine Workshop 2004.
  8. Student Associate, CET, IIT Kharagpur (Jan-Jun 2002)

    Implemented a "Digital Simulation Software" using JAVA and XML, for remote digital circuit lab experiments. Implemented basic logic gates (74XX series) and wiring algorithms for creating and simulating both ICs and complex digital circuits. Circuits designed at the client side, were simulated on the remote server using RMI, after proper user authentication.


  1. PDA based Teleconsultation over Wireless LAN

    J.Mukherjee, A.K.Majumdar, J. N. Maiti, Suman Kundu, Sumit Kumar, V. Pallavi
    Proceedings of 2nd APT Telemedicine Workshop 2004 on "Wireless Communication for Health"


Personal Summary


  1. Predictive Customer Service Environment - US Patent Application 20120076283
  2. PxHelper: a web based guidance system
  3. Article on The Hindu Businessline covering WiredGurus Challenge 2005
  4. DADP Mobile Application Development Contest 2003
  5. Other references available on request
Last Updated: Jan 23 2018