1. VARUN SUBRAMANIAN
Ph.: 540 998 4592 Email: varuns92@vt.edu
1221 University city Blvd, Blacksburg, VA 24060
LinkedIn: https://www.linkedin.com/in/varun-subramanian
OBJECTIVE
Seeking a full time position in the area of Software development.
ACADEMIC DETAILS
M.S. Computer Science Virginia Tech GPA 3.60 Aug 2015- May 2017
Graduate Courses: Operating Systems, Multiprocessor Programming, Advanced Parallel
Computation, Theory of Algorithms, Virtual Environments, Cloud Computing, Advanced Machine
Learning, Information Visualization, Software Engineering
B.E Electrical & Electronics, PSG Tech , India CGPA:8.54 Jun 2009 - May 2013
TECHNICAL SKILLS
Programming : C, C++, Python, CUDA, OpenMP, MPI
Operating Systems : Linux, Windows 7/10
Tools : GDB, LLDB, WinDbg, Perforce, Git
RELEVANT EXPERIENCE
System Design Engineer, Nvidia May 2013 –July 2015
Responsible for power characterization of GPU’s for App note measurement and successfully
performed the characterization for Kepler and Maxwell Notebook GPU’s.
Worked on Software and feature Validation of tools intended for power measurement.
Implemented systems and changes in Methodology that led to better App note measurements
and more rails being brought under characterization.
Researched the methods to improve the power efficiency of GPU’s by analyzing the leakages,
dynamic power consumption of memory and other non-core parts that resulted in reducing power
consumption for the next generation micro-architecture.
System Software Intern, National Instruments May 2016 –Aug 2016
Implemented the Asynchronous device Initialization of device drivers in NI- APAL layer for
Windows platform that provided the layer with parallel driver initialization feature that lead to
significant reduction in installation times for a group of drivers. The project involved
understanding of Windows Driver Model (WDM) and implementing IRP handling
asynchronously to achieve parallel initialization.
Ported NI- VISA (Virtual Instrumentation Software Architecture) server application (Mac OS)
from legacy carbon framework to Cocoa Framework and also implemented systems to ensure
that the application does not undergo AppNap.
PROJECTS
Kernel Design
Designed a kernel, which involved implementing, various scheduling mechanisms, User program
support, Creating a new file system for the kernel, System Calls Design and support for most of Unix
system-calls for Pintos, a Unix like Operating System.
Virtual memory Design
Designed and implemented a complete virtual memory system with design of frame table,
Supplemental page table, and swap system for Pintos, a Unix like kernel.
Alternative approach to ceilometer monitoring
Implemented a Machine Learning based approach with SVM and ARIMA modelling to reduce the data
footprint of the telemetry component of OpenStack, Ceilometer leading to an effective polling of the
resource utilizations of the Virtual Machines.