Swetha Varadarajan

Fort Collins, CO *****

720-***-**** Email: aczvs8@r.postjobfree.com Linkedin: https://in.linkedin.com/in/swesri Available: Jan 2017

WORK EXPERIENCE

Colorado State University, Fort Collins, CO

Graduate Teaching Assistant- Parallel Programming Aug 2016 – Present

Set up HTML scripts for the course webpage. Python/Shell scripting for automatic testing and grading of programming assignments through the web interface. Prepared quiz questions for the class.

Worked on 8-core single multi-core Intel Sandy bridge for OpenMP programs, Cray Model XE6 with 2688 cores for MPI programs and NVIDIA Maxwell architecture for CUDA programs. Colorado State University, Fort Collins, CO

Graduate Research Assistant – High Performance Computing Aug 2014 – July 2016

(Masters Thesis) Parallelization and optimization of computations found in RNA-RNA interaction application. Re- writing polyhedral computations in a Domain-specific language called AlphaZ that generates optimized OpenMP C code with user specified program transformations. Projected to improve performance from O(N6) to O(N5).

(Mentor) for a class project during Fall 2015. Supervised and helped a student to improve the parallel performance of

“McCaskill algorithm” from O(N4) to O(N3) through program transformations.

Micro-benchmarking on GPU: Performed benchmarking on several kernels similar to Matrix Maultiplication on 3 different GPU architecture to measure and tune performance (GFLOPS/s) and bandwidth(Gbytes/s)

Micro-benchmarking on CPU: Performed benchmarking on kernels like Triangular Matrix Mulitplication and other higher data dimensional kernels (6 dimensions) on Intel Haswell to measure and tune performance (GFLOPS/s). Studied vectorization(SIMD),optimization reports and also the effect of various compiler flags on ICC and GCC.

Performance tuning based on tile size selection on a given CPU architecture and for a problem size. Performed an exhaustive search of tile sizes. Surveyed various literature that uses Machine learning algorithms for this purpose.

(Research paper) A regular weekly study/review of research papaers in the field of High Performance Computing. The popular ones would be, the Roofline model and 12 ways to fool the masses when reporting performance results. Responsible for maintaining the research group(CSU Melange) website that is been setup using DokuWiki. Colorado State University, Fort Collins, CO

Research Programmer - Department of Forest, Rangeland and Watershed Stewardship Summer 2014

Studied and analyzed the given CUDA application that has been implemented for calculating Haralick texture features

Modified certain files/functions so that it can be adopted for geo-referencing features and algorithms. Lucas-TVS, Chennai, India

Intern – Embedded systems laboratory Dec 2012 – April 2013

Developed a Simulink-Stateflow model for automatic stop-start mechanism of modern cars.

Generated the code for TI’s TMS320F controller series using Embedded Coder and TI’s Code Composer Studio

Developed a PCB design for input and output signal conditioning. Simulated signals through switches, function generators and ran laboratory level experimentation for TMS320F series. Results were viewed in oscilloscope.

Extended the model to incorporate the Ultra-capacitor based Regenerative Braking system too. TECHNICAL KNOWLEDGE

Languages: Expert- C,Python, C++,CUDA,MPI,OpenMP,Shell script,ASM 85/86 Intermediate- HTML,VHDL,MIPS Novice- Java Systems: Expert- Ubuntu, Fedora, Windows 7 Intermediate- Mac Novice- Unix Architecture:Intel CPU- Haswell,Ivy-bridge,Sandy-bridge NVIDIA GPU- Maxwell, Kepler, Tesla Software: Eclipse, Nsight, Simulink, Matlab, Vtune Amplifier, Perf, CACTI, Gem5, Noxim, CCS, LATeX and MS Office EDUCATION

Colorado State University, Fort Collins, CO Aug 2013 – Present Candidate for a Master of Science in Electrical Engineering, GPA: 3.55/4 Expected Graduation: Spring 2017 Related Courses: Parallel Programming, Parallel Processing, Hardware/Software co-design of Embedded systems, Computer architecture, Machine Learning, Linear Programming & Network flows Academic Projects:

Extensive case study on NP-Complete Problem (0/1 Knapsack solver). Implemented the naïve approach O(NC) memory, a memory efficient version O(C) and the parallelization of it in CPU and GPU(CUDA).

Developed a mapping and routing solutions for a Given an MPSoC of heterogeneous cores (with 2 types of cores) and a task graph, taking chip wide power performance constraint in C language.

Used Reinforcement learning to control the motion of a marble (has mass, real-valued position and velocity) on a one-dimensional track using Python.

Implemented clustering algorithms (K-means, Hierarchical, SOM, EM) in Python and compared its effectiveness and usefulness with respect to size & type of data-set, number of clusters and also with the existing software versions. SASTRA University, India

Bachelor of Technology (Honours) in Electronics & Communcication Engineering May 2013 Related Courses: Micro-Processors, Micro-Controllers, VLSI Design, Embeded Systems, Object Oriented Programming(C Digital Electronics, Electronic Circuits, Software Testing, HDL programming, Linear Integrated Circuits. PUBLICATIONS

Gyratory assisted Info Hide a Nibble differencing for Message embedding, Information Technology Journal, 2014, ISSN 1812-5638/ DOI:10.3923/itj.2014.2005.2010

Captivating CODEC Stego (CCS): A Cover on Camouflage,Research journal of Information Technology, 2013, ISSN 1815-7432/DOI:10.3923/rjit.2013.160.170

INTERESTS / ACTIVITIES

Teaching, Formulating quiz/trivia/challenging questions for events/competitions.

Hiking, Musical Instrument(Veena), Vocal(Indian classical Carnatic Music).