1235 Wildwood Ave, Apt #1
Sunnyvale, CA 94089
USA
Email: ajitdatar@gmail.com
URL: http://ajitomatix.com
To work in a research and/or development position in the Computer Science Field.
Summer internship at Microsoft IDC [ June 2004-August 2004 ]
MTS (Software) at Persistent Systems Pvt Ltd, India [ March 2003-August 2003 ]
Software Developer at Zensar Technologies, India [ October 2002-March 2003 ]
Overall GPA: 3.8 out of 4.0
Subjects: Advanced Machine Learning, Advanced Computational Logic, Digital Signal Processing, VLSI, Digital Image Processing, Natural Language Processing, Advanced Computer Architecture.
Thesis - Generating hyperbolic patterns for regular and non-regular p-gons - Improved and implemented algorithms to generate hyperbolic designs based on the work of Dr. Douglas Dunham. Developed HyperArt -- a C++/Qt based cross-platform programming framework and application to experiment with repeating hyperbolic tessellations. See http://ajitomatix.com/Projects/HyperArt and http://hyperart.sourceforge.net for details.
Essay grading and Analysis Logic - Designed, developed and deployed EGAL -- a system to grade essays based on four criteria - Gibberish Detection, Relevance assessment, Factual statement identification and Factual accuracy. Provides a CGI/Perl web interface and Perl modules for the four criteria and an installer. Implemented the factual accuracy module. Responsible for overall design, coordination and deployment. The implementation details and sourcecode is available at http://sourceforge.net/projects/egal.
Fast Multiresolution Image Querying - FMIQ is an implementation of a paper of the same title. Developed in python, it is a search technique which uses Haar wavelet transform to compute signatures of the images in its database. These are then used to do a image query by example or by a rough sketch. Complete source code is available at http://ajitomatix.com/Projects/FMIQ
Subjects: Data Structures and Algorithms, Computer Graphics, Databases, Operating systems, Computer Networks, Compilers, Object Oriented Design, Software Engineering.
Designed and implemented an Optical Mark Recognition system to grade multiple-choice style questions. The system placed minimum restrictions on the scanned document.
Experimented with Devanagari Optical Charater Recognition system using a 3-layer, fully connected backpropagation neural network. Successfully converted scanned Devnagari text to electronic format with an accuracy as high as 90%. Both the projects used C++. MFC was used for the GUI.
Computer Languages: C, C++, Python, Perl, XML, HTML.
Compilers/Frameworks/Tools: Qt, MFC/Win 32, GNU Dev tools, cvs.
Platforms: Windows 98, Windows XP, Windows 2003 Server, Linux, Solaris.
Graphics Programming: OpenGL.
Latest version of this document can be obtained from http://ajitomatix.com/Resume.
References available upon request.
Last modified Nov 2005.