1 Shounak Roychowdhury, Ph.D
Total Page:16
File Type:pdf, Size:1020Kb
Shounak Roychowdhury, Ph.D. 10213 Prism Dr., Austin, TX, 78726 || 650-504-8365 || email: [email protected] Profile • Software development and research experience at Oracle and LG Electronics. • Deep understanding of data science methods: machine learning; probability and statistics. • 5 US patents and 40+ peer reviewed publications in international conferences and top refereed journals Research Interests • Published research papers on computational intelligence, neural networks and fuzzy theory, numerical optimization, and natural language processing, and information theory. Education • Ph.D. (Computer Engineering), University of Texas at Austin, Austin, TX, (Dec. 2013) o Dissertation: A Mixed Approach to Spectrum-based Fault Localization Using Information Theoretic Foundations. (Machine Learning in Software Engineering) • M.S. (Computer Science), University of Tulsa, Tulsa, OK, (May 1997) o Thesis: Encoding and Decoding of Fuzzy Rules Patents • Chaos washing systems and a method of washing thereof (US Patent #5,560,230) • System and method for generating fuzzy decision trees (US Patent #7,197,504) • Method for extracting association rules from transactions in a database (U.S. Patent # 7,370,033) • Expediting K-means cluster analysis data mining using subsample elimination preprocessing (U.S. Patent # 8,229,876) • Bayes-like classifier with fuzzy likelihood (U.S. Patent # 8,229,875) Computer Languages • Python, Java, C/C++, MATLAB, R, SQL, PL/SQL, Perl, Ruby, Tcl/Tk Teaching Experience Adjunct Faculty Texas State University 2017- Present Professional Experience Hewlett Packard Enterprise, Austin, TX (Oct 2018 - present) Expert Technologist • Executed software development processes for composable rack team of HPE’s OneView cloud management system. • Developed a Python-based system to test the scalability of OneView connections across multiple layers of Plexxi switches. • Scripted Python codes to analyze log files of OneView using granular representation of Bag-of-Words technique. Oracle Corporation, Redwood Shores, CA and Austin, TX (Mar 1997-Sep 2018) Principal Member of Technical Staff (Oracle Database and Migration Tools) (2009-present) [Database Cloud Services] • Amazon Redshift data migration to Oracle DataWarehouse Cloud • Integration of RUEI and Oracle Enterprise Manager (OEM) in the context of Database Schema Cloud Services. • Designed and coded Oracle VM statistics collection and reporting system with RESTful APIs (Jersey 2.0 and Apache Http) and Apex (Oracle Restful Services), 1 • Designed and coded orasql plugins in JRuby for Logstash, Elasticsearch and Kibana (ELK) for log analytics of Oracle Cloud. • Automated log analyzer of event logs, syslogs; and patent proposal of security mechanism for log data. • Analysis of database audit logs, correlation analysis of database instance metric logs. [Database Tools] Designed and implemented the following features for 4 major releases of SQL Developer (2.5, 3.0, 3.1, 4.0, 4.1) • Data masking and redaction feature of Oracle 12c R1. The feature was very well recognized at Oracle Open World-2013. • Coded and designed pan based MapViewer using Oracle DB’s Spatial data capabilities (Oracle 11g R1) for geospatial users, similar to Google Maps. • Coded Oracle Scheduler feature (Oracle 11g R1 and Oracle 11g R2), replacement of DBMS Jobs feature. Implemented in Java 5.0 Swing. • Enhanced and fixed bugs of the several command line features of SQLcl. [Machine Learning Research] • Design and implemented data clustering algorithms for Oracle Spatial data. Improved K-means clustering algorithm by improving the discovery of better initialization points through sub-sampling algorithm. US patents on the algorithm. • Improved on the existing classifier design for spatial data clustering. • Design of optimal clusters using neighborhood graphs (MATLAB). • Design of uncertain-weighted clusters using association rules from large data sets (MATLAB) • Designed feature-based clustering technique using projective maximal stochastic information (MATLAB) • Designed feature removal algorithm using hierarchical granularity of proximal data. • Proposed of collaborative co-clustering using granular models. • Simulation of drift of regression model in long term datasets. • Word sense disambiguation algorithm using web sources for natural language processing • Extraction of word clusters based on word frequency and phrase frequency from textual data • Proposal of private key generation from multiple public keys using compressed private key index cluster Senior Member of Technical Staff (Oracle Data Warehousing Group) (2001-2008) Worked on major features of 8 releases of Oracle Warehouse Builder (OWB) • Designed several functional and design specifications for dimension modeling of different versions and subcomponents (e.g. Tables, Views, Materialized Views, Dimension, Time Dimension, Cube, Maps, Models, Validation and Compilation units) of OWB till OWB11gR1 (2008). • Implemented OWB’s Dimensional API layers for Star and Snowflake Schema in Java/XML and Swing UI components (dimension, Time dimension, cube, mappings, models, and validation and compilation units). • Developed dimensional and relational portions of command line scripting language called OMB+ (in Tcl/Tk). • Designed an extensible OMB+ testing framework (Tcl/Tk) and Html. Over the years it became the main testing framework to test OWB infrastructure at the API level. • Implemented entire workflow of the time dimension; from the design and implementation of the related UI wizards (Java) to time data generators (PL/SQL) for fiscal and calendar years. Maintained generators for time data and its migration across releases. • Performance evaluation of Rational ROSE model changes for various releases and evaluate its effects on the performance of the new APIs. Also minimized the rate of JDBC calls and SQL statements to the metadata repository. • Improved persistence of Java objects in OWB metadata repository using Oracle Toplink. Member of Technical Staff (Oracle Tools Group) (1997-2001) • Designed and developed Oracle 8 features for Query Builder/Browser of Developer/2000; Redesigned Query Builder Parser for Object-Oriented capabilities of Oracle 8. Implemented parser in LEX/YACC using C. Proposed computational methods for designing fuzzy queries. • Developed test plans for backend testing as well as front-end testing of Query Builder. Built and maintained bug fixes and stage of Browser/Query Builder for different releases. Fixed Query Builder/Browser backend bugs; Developed regression test suite for Win16 and enhancements of Win32; Developed Query Builder's regression system (OBTEST); Enhanced BENT (internal win32-based Version Control system) 2 • Developed a CGI-bin based Load Balancer for Oracle Forms and Oracle Application Web Server 2.0.4. University of Tulsa, Tulsa, OK (Jan 1995 – Feb 1997) Graduate Research Assistant : Department of Computer Sciences • Applications of fuzzy sets and rough sets in database mining on relational databases. Involved in design and development of rules using principles of data clustering, rule induction, rule clustering and knowledge compression; derived heuristic knowledge in form associative induced fuzzy rules. • Data security and privacy for mined rules from databases. LG Electronics Inc. Central Research Laboratory, Seoul, South Korea (Jan 1992 – Dec 1994) Research Engineer [Neural Networks, Fuzzy Systems, and Chaos Group] • Proposed new theoretical foundations on triangular-operators for fuzzy systems; developed new defuzzification methodologies for fuzzy controllers; worked on the theoretical development • Applications of Hebbian learning, supervised and competitive learning, radial basis fuzzy networks in consumer products like washing machines, air-conditioners, intelligent machines and embedded software designed and implemented a MS- Windows (Windows 3.1/SDK) and XMotif/Xview (UNIX) software for fuzzy control simulator. • Designed for Chaos Washing Machine whose prototype was shown at IFSA conference 1994. Korean TV Broadcasting Station (KBS1) aired my work on fuzzy systems at LG research center in May 1994. [Image Processing and Compression Group] • Researched holographic image processing and formation of moiré images (optical and holographic) for non-destructive surface regeneration and reconstruction with phase-shift methods; image generation/image reconstruction, image filters, edge detection and image enhancement; image compressions with RLE and DCT were also developed; developed and implemented image processing software packages (called analyzers) for filtering, compressing, decompressing, and segmentation (using FFT & DCT) of speckled/moiré images produced by holograms and optical systems respectively in Borland-C on DOS system (GoldStar 486). TCIL-BellSouth Ltd. New Delhi, India (Aug 1991 – Dec 1991) Software Engineer • Developed a part of the GUI front-end for large telecom software using Intergraph's graphic language. Indian Institute of Science, Bangalore, India, (Aug 1990 – May 1991) Research Assistant • Development of a simulation packages of stochastic Petri nets and Markov models. Publications Book Chapters 1. S. Roychowdhury and J. Ghosh, “Robust Laplacian Eigenmaps Using Global Information, Manifold Learning Theory and Applications, Editor(s): Yunqian Ma, Yun Fu, CRC Publication, 2011. 2. S. Roychowdhury, “A survey on the theory of defuzzification,” Handbook of Granular Computing, Eds. W. Pedrycz, A. Skowron, and V. Kreinovich, 2006. 3. S. Roychowdhury, “An inquiry into the theory of