Education

  • Ph.D : University of Arizona (Aug 2014 - Present)
    Advisor : John H. Hartman
  • MS : University of Arizona (Aug 2012 - May 2014)
    Advisor : John H. Hartman
  • Bachelor of Engineering : Hankuk University of Foreign Studies (Mar 2001 - Jun 2009)
    Advisor : Sangyoung Cho

Work Experience

  • Minigate (Sep 2010 - Jul 2012)
    Role : WAC & W3C mobile web widget framework research and development
  • Entaz (Feb 2007 - Feb 2009)
    Role : Mobile game framework and webbrowser development
  • Ensi (Aug 2005 - Feb 2007)
    Role : POS system software development
  • AT Solution (Jun 2004 - Aug 2005)
    Role : Mobile (non-smartphone) application development

Download CV (PDF, last updated at Oct-26-2015)

Research Interests

Operating Systems, Distributed Computing, Cloud Systems and Bioinformatics.

Research Papers

  • H-Synthesizer: Analyzing Large-scale Sequence Data In The Cloud (MS Thesis)
    PDF version

Research Projects

  • Syndicate: CDN-backed Cloud Storage System. A cloud storage system that used CDN and cloud technologies, and developed by Jude Nelson. Responsible for implementing domain specific gateways such as AG for iRODS and UG for Hadoop. Developed AG disk-polling driver, UG-IPC (TCP/IP) and UG-Node.JS (HTTP). These work inspired in design of Stargate, H-Syndicate, iPlant Border Message Server and H-iRODS. Used C, Python, Javascript, NodeJS.
    Syndicate project page, UG-Node.JS project page
  • Stargate: Inter-cluster Storage Integration System. A system for integrating remote data storage systems and running Hadoop computations against. To provide efficient data transfer between remote clusters, Stargate used recipe based data transfer and various caches (web cache/cluster cache) smartly. Stargate provided Hadoop filesystem interface and scheduled transfer order so that Hadoop computations worked efficiently against data at remote clusters. Used Java, Hadoop.
    Project page
  • H-Syndicate: Hadoop FileSystem implementation of Syndicate. A software for integrating Syndicate cloud filesystem and Hadoop. This software enabled Hadoop computations to be able to access Syndicate Volume, and provides efficient access to data by controlling task assignment of Hadoop using the presence and locations of Syndicate local cache. Used Java, C, Hadoop.
    Project page
  • Kogiri: Massive Comparative Metagenomics Analytic Tool. A tool for analyzing large NGS metagenomics samples by pair-wise k-mer comparison fashion. Kogiri used Hadoop MapReduce to analyze large number of samples, each in gigabyte size. Kogiri successfully analyzed 125 samples, 551GB in total size of TARA Oceans Viromes successfully. Used Java, Hadoop MapReduce.
    Project pageMetagenomic research (HurwitzLab)Conference abstract(2015 Aquatic Sciences Meeting)
  • BioSpectra: Metagenomic Read Classification System. A classification system for metagenomics reads using k-mer. BioSpectra allowed k-mer searching against an index of k-mers from known reference sequences, scoring and classifying using Lucene search engine. Used Java, Lucene.
    Project page
  • iPlant Border Message Server: iPlant Data Store Event Notification System. An event notification server system of iPlant Data Store. All filesystem events genereated by iPlant Data Store are transferred to this system then, classified and routed to users. Used C, RabbitMQ.
    Project page
  • iRODS-FUSE (4.1, 4.2): A FUSE Client of iRODS FileSystem. iRODS-FUSE client allowed users to mount iRODS storage onto local directory hierarchy. It was re-designed from scratch due to issues at inconsistent content and cache management of existing iRODS-FUSE (3.2). The new design solved these issues and provided better read and write performance (26% reduction of read time and 88% of write time compared to 3.2). The new implementation was accepted by iRODS consortium and contained from iRODS 4.1.4 release. Further enhancement for read and write performance (68% reduction of read time and 91% of write time compared to 3.2) also was accepted for iRODS 4.2 release. Used C.
    Project page (iRODS FUSE Development)Project page (iRODS)
  • H-iRODS: Hadoop FileSystem implementation of iRODS. A software for integrating iRODS and Hadoop. This software enabled Hadoop computations to be able to access iRODS. This software was also designed to work at a situation that iRODS and Hadoop clusters are remote and connected through WAN. To efficiently handle large intermediate outputs of MapReduce computation, this software used HDFS as a temporary storage and transfer only final dataset to iRODS system through WAN. Used Java, Hadoop.
    Project page

Timeline