Ujan Sengupta has a Master's degree in Computer Engineering from North Carolina State University and a Bachelor's degree in Electronics and Communication Engineering from Manipal Institute of Technology. He has skills in programming languages like C/C++, C#, Swift, LLVM, and tools like IBM Watson, Bluemix, Xamarin. He has internship experience at IBM and Philips India developing mobile applications and using technologies like Watson APIs, Xamarin, and C#. His projects include computer vision algorithms, modifying a GPGPU simulator, developing architecture simulators in C/C++, and implementing optimizations and transformations in the LLVM compiler.

1. Ujan Sengupta
Raleigh, NC • (919) 995-0174 • usengup@ncsu.edu • LinkedIn • GitHub
Education
 North Carolina State University May 2017
- Master of Science in Computer Engineering (GPA - 3.61/4.0)
- Courses: Computer Design and Technology, Code Generation and Optimization, Architecture of Parallel Computers,
Computer Vision, GPU Architecture, Design and Analysis of Algorithms, Computer Graphics
 Manipal Institute of Technology May 2015
- Bachelor of Technology in Electronics and Communication Engineering
Skills
 Programming Languages: C/C++, C#, Swift, LLVM, Android, Embedded C, CUDA, Python, Java, OpenMP
 Tools: IBM Watson, Bluemix, Xamarin, Flex/Bison, Git, MATLAB, XCode, Eclipse, Arduino, Image File System
Experience
 Software Development Intern - IBM, Research Triangle Park (North Carolina) Aug, 2016 - Present
- Currently developing a proof-of-concept application with the Sales Model Transformation team on Territory
Optimization and Planning using Watson APIs
 Software Development Intern - IBM, Research Triangle Park (North Carolina) Aug, 2016
- Developed a cognitive chat application leveraging IBM Watson technologies – Native iOS (Swift 2.2)
- Used Watson APIs hosted on Bluemix to analyze the chat data in real-time; the analysis provided the conversational tone
of individual users and real-time keywords were generated to aid in contextual tagging of the entire chat
 Research Intern - Philips India Ltd., Bangalore (India) Jun, 2015
- Developed an application which enabled clinicians to monitor real-time patient data on mobile devices
- Made use of Xamarin (specifically the Xamarin.Forms API) to develop a shared code-base between the iOS and Android
versions of the app thus reducing code length, time to market and resources required to develop updates – C#
 Research Intern - Fraunhofer FKIE, Wachtberg (Germany) Aug, 2014
- Developed an algorithm to study CBRNE (Chemical, Biological, Radiological, Nuclear and Explosive) sensor data from
unmanned NATO surveillance robots. Also wrote virtual drivers for the Robot Operating System (ROS) framework to
provide test scenarios for simulated hazardous conditions – Python
Projects
 Computer Vision: Performed following analyses and experimentation on images using C++ and IFS:
- Application of a simulated-annealing variation of the diffusion equation on noisy images; the objective was to study the
extent of edge-preserved smoothening that could be achieved via this technique.
- An approach to find the center of circles in images; I used an algorithm leveraging the Generalized Hough Transform
(GHT) and gradient descent.
- Stitching multiple images by finding correspondences between them; this was done both manually and using various
interest-point detectors and local area descriptors like Harris-Laplace, SIFT, etc. and homography matrices.
- Currently involved in a collaborative project among the History and ECE departments at NCSU. The objective is to
automatically generate visual narratives using images of historical events by training an RNN, using an exhaustive data-
set.
 GPGPU Simulator: Modified the GPGPU Simulator to implement Criticality Aware Warp Acceleration (CAWA) and
successfully attempted to replicate the results of the paper titled “CAWA: Coordinated Warp Scheduling and Cache
Prioritization for Critical Warp Acceleration of GPGPU Workloads” by Lee et al – C++
 Architecture Simulators: Developed the following architecture simulators in C/C++ and used them to measure and analyze
the performance for various configurations in each context
- A cache simulator consisting of an L1 cache augmented with victim and L2 caches.
- Extended the above simulator to support SMP systems by implementing bus based cache coherence protocols like MSI,
MESI, MOESI and Dragon.
- A hybrid (bimodal + gshare) branch predictor simulator supported by a Branch Target Buffer.
- A dynamic instruction scheduling simulator using the Tomasulo algorithm for superscalar pipelines.
 LLVM Code Generator and Optimizer:
- Implemented LLVM libraries to perform various optimizations (DCE, CSE, GCM, LICM etc.) and evaluated their impact on
the vital statistics of a program – C++
- Implemented an LLVM library to perform a vectorizing transformation to exploit Superword Level Parallelism and created
an LLVM generator for a C-like language – C++