Prem Janardhan

Prem Janardhan

(H) 732-***-****, (Mob.) 848-***-****, ****.*********@*****.***

Summary

Over 15 years of Software Development and Information Technology experience starting with 7 years in algorithm research and development for medical products followed by over 8 years in software design and development for commercial applications.

Skills include high proficiency in C, C++, Java, C#, SQL (queries, triggers, stored procedures, functions) on UNIX and Windows. Application experience includes Client-Server computing and N-tier Web Development using Java and ASP.Net on Oracle AS, Tomcat, Microsoft IIS etc. servers.

Domain knowledge includes Healthcare, Banking, eCommerce and Telecommunications and experience in the complete software life cycle. Primary strength is effective problem solving and communication skills.

Education

Carnegie Mellon University, Pittsburgh, USA (1996–1999).

PhD ABD, Computer Science

Carnegie Mellon University, Pittsburgh, USA (1993–1996).

MS, Computer Science, 4.0 GPA

Courses in Algorithms, Programming Language theory, Artificial Intelligence, Operating Systems, Computer Graphics, Image Processing, Computer Vision, and Parallel Computing,

Susquehanna University, PA, USA (1989 – 1993).

BA, Computer Science and Mathematics, University Honours, summa cum laude, 4.0 GPA, international student (tuition, room and board) scholarship, valedictorian

Work Experience

AT&T Middletown, NJ (Apr 2008–Apr 2009).

Responsible for test plan development and execution for RuBY(Rules U Build Yourself) – a rule-based system for automatic routing and handling of anomalous events occurring on AT&T Telecommunication equipment within their Uverse (IP-based internet, voice, TV and video on demand) service. RuBY programs an expert system to handle reported anomalies for resolution by routing them appropriately to a service technician, or self healing. Is an integral part of maintaining (SLAs) Service Level Agreements in the AT&T Communications network.

ETA Infotech, Dubai, UAE (Mar 2005–Dec 2007) – Manager Systems

First Gulf Bank, Abu Dhabi, Online Certification System

The Online Certification System is a web portal application used for testing Bank employees in order to upgrade their job related skills as well as provide an evaluation mechanism for deciding on motivational incentives. The portal was developed on the MS SQL Server platform using C# ASP.NET. Convenient user interfaces for administrative purposes like uploading from MS-Excel etc. were also provided. I was responsible for the complete software cycle from conception through development to implementation.

Mediview / Gastroview / Bronchoview

This is a Digital Image Management and Reporting solution that permits doctors using a medical imaging hardware to store and manage images based patient records. It has a flexible interface that allows integration with any Hospital Information System (HIS). Features include full motion video capture through (e.g. Avermedia) grabber cards; video editing suite to permit capturing stills from streams, and exporting them to PowerPoint for presentations; and a flexible reporting module that allows creation and customization of reports from predefined templates. I was involved in maintaining the code and creating customer centric variations including implementation at client sites. The platform used was Windows Visual C++ DirectX, on MS SQL Server).

ETA Infotech & Builder's Portal, Dubai, UAE (Sep 2000–Mar 2005) – Manager Systems

ATM (Automatic Teller Machine) Controller

The Controller is a multithreaded ATM Host Program to interface between a physical ATM and the Banking software at the server. Although designed for an Oracle based Banking application, the interfaces were kept flexible for use with several types of ATMs and Banking applications. I was responsible for the complete development and implementation at third world country sites.

BuildersV2B.Com

BuildersV2B is the largest private sector construction procurement portal in the UAE, implemented originally in Java/JSP/J2EE and later re-implemented in ASP.NET with facilities for construction contracting buyers and construction material suppliers to posting purchase enquiries, respond with price quotations, conduct forward and reverse auctions, issue purchase orders, post cheque and payment information and perform related administrative functions. I was the head of a 7 member team that developed and implemented the entire project and was the principle architect and system integrator. It provided on line buyer-vendor interaction and had a few thousand registered members.

Cellomics Inc., Pittsburgh, USA (Sep 1999-Sep 2000). – Imaging Algorithmist

Cellomics Array-Scan is an instrument for high-speed acquisition of images from a micro-array of cell cultures. The images are processed by quantitation algorithms. Neurites are linear processes extruded by neurons in the process of their growth. Stimulating their growth is crucial in study and treatment of neural problems such as Alzheimer’s. Various compounds were tested against neuron cultures. The resulting growth had to be robustly quantitated by the algorithm, to account for biological and imaging variation. I worked with biologists to construct and code a multithreaded algorithm in C++ with an adequate noise rejection rate. A patent has been issued for this work: US Patent 6,986,993, A System for Cell-based Screening, Ghosh RN, DeBiasio R, Janardhan P, Published 1/17/06.

Computer Vision Lab, University of Tokyo (June 1996-July 1997) Visiting Researcher,

Recognizing Human Actions from Infrared

Human actions were captured in silhouette using an IR camera. An open-source HMM system was trained to recognize actions using these images. Curvature and intensity were analyzed along the outlines of these images were analyzed using a Hidden Markov Model to robustly identify the individual actions. This was entirely an individual effort.

Hitachi Production Engineering Research Lab, Yokohama, Japan (1995, 2001-2002),

Visiting Researcher.

Industrial Inspection using Images

Laminography is an X-ray imaging technology that keeps a sensor fixed and rotates the sample to generate a cone-shaped image that is reinforced only in the desired focal plane. It generates shadow artifacts due to the imaging process. This project involved creating a basic physics model of the X-ray laminography imaging process, and numerical reconstruction of the original target from the obtained images, in an effort to eliminate shadow artifacts, and also improve the performance of the machine inspection system. The artifacts occur because of shadowing of one part of the sample by others in another 3D plane (i.e. focal separation is not perfect due to the way the technique works). I later returned to the same lab to assist with improvements in their machine inspection system. Coding was done in C/C++ and their environment was Solaris.

Research Experience

C.M.U. Computer Science Department, Pittsburgh, (Sep 1993-Sep 1999), PhD Researcher

Drosophila Embryogenesis

System to recognize the distinctive stages of Drosophila Embryogenesis from optical section images of embryos injected with appropriate fluorescence agents without disturbing the embryo’s development. Image Understanding Algorithms were applied to the generated images to identify the stages of development automatically. The identified stages were used to generate a coordinate framework whereby we could compare different embryos to identify anomalies and characterize their development. The image collection was performed by members of the Minden Lab situated in the Biology Department. I developed the algorithms, and wrote the code for the system. I also presented the results at several technical conferences: Computer Vision (CVPR) Workshop on Biomedical Image Analysis, and also at a Japanese Computer Vision conference. The environment used was C++ under Unix using some ready-made image handling / processing libraries (Vista) and user interface building systems (Xforms).

Tracking of Apoptotic Cells:

The objective of this system was to track apoptotic cells (excess cells pruned by the organism’s own developmental process) from images treated with fluorescent dye for visualization. This system was used to identify how many cells and which ones died at various stages of development. All the algorithm development and coding was personally undertaken by me using C++, XForms and Vista under a Unix environment.

Parallel Connected Components on iWarp:

Implementation of a parallel Connected-Components algorithm in a parallel dialect of Fortran (Fx) on the iWarp computer which has 64 nodes, and testing of its performance. Connected Components is a very common algorithm that is used in image processing that helps to label distinct image segments. The entire implementation was my responsibility.

Robot Navigation and Map-Building:

Designing and implementing a geometric algorithm to utilize output from Sonar mounted on a wandering robot, to build a map of a maze, and then continue to navigate the maze successfully to locate specific objects within it. Xavier, a CMU Research Robot, was programmed to perform this navigation. I was part of a team of 3 individuals who did this work. The algorithm used readings in perpendicular directions to orient itself with respect to surrounding walls and identify directions in which it could move, whether it was in a corner etc.

Deformable Templates for Face Recognition:

Deformable templates are pseudo-mechanical algorithmic widgets that are designed to conform to particular kinds of patterns in images by treating the image as an energy surface; the objective of this system was to develop a Deformable Template algorithm for the recognition of eyes, lips and eyebrows. Various different types of templates were used for each face part. This was a class project and performed entirely by me using C++ / Xforms and Vista under Unix (Mach).

ARPA Image Understanding Video 1994:

The Image Understanding research community annually submits image and video results to be collated together into a presentation to show what the research output was that year. I assisted Dr. Takeo Kanade, the director of the Carnegie Mellon University Robotics Institute in creating this video for 1994. I prepared a script, collated the inputs from various research groups, decided what data to include, set up the voiceover and otherwise assisted in production of the video.

Contact this candidate