Prakalp Somawanshi
Email: prakalp.somawanshi@gmail.com
C, CUDA, MPI, OpenMP ( 4 years Experience in HPC )
Mobile : +91- 9881 908 059
SUMMARY :
 C, CUDA developer with good problem solving and analytical skills.
 Expertise in design, analysis and optimization of parallel algorithms.
 M.Tech in Control and Computing from IIT Bombay.
 Ability to adapt efficiently to the changing environment.
ACADEMICS:
 M.Tech in Electrical Engineering (Control & Computing), Indian Institute of Technology Bombay, Mumbai.
August 2005 - June 2008. [CGPA 8.7 / 10 ]
 B.E. in Instrumentation & Control. University of Pune, June 1999 - May 2003. [ 70% , First class with
Distinction]
SKILL SETS:
 Programming Languages: C, CUDA-C, C++, OpenMP, MPI, OpenCL.
 Debugging and Analysis tools: CUDA-gdb, CUDA-profiler, MPI Trace, SCALSCA, Intel Vtune., VAMPIR.
 Scripting Languages: Bash, Perl.
 Numerical Packages & Libraries: BLAS, NTL, MATLAB, Scilab.
 IO Format : NETCDF, HDF, Parallel NetCDF.
 Domain Applications : WRF (Weather ), OpenFOAM(CFD), Option Pricing – European & American
(Finance).
PUBLICATIONS:
 Prakalp Somawanshi, Animesh Agarwal, Mahesh Khadatare, Shrirang Karandikar, Dr. Chinmoy Bhatt.
“Accelerated SAR image generation on GPGPU”. Asia-Pacific international conference on Synthetic
Aperture Radar, Seoul, South Korea, September 2011.
 Prakalp Somawanshi, Tejaswi Kumar, Mahesh Khadatare. “Acceleration of Long-Wave Rapid Radioactive
Transfer Model on GPGPU”. GPU Technology Conference 2011, San Diego, CA.
 Narayana Darapaneni, Prakalp Somawanshi, Meghana Joshi. “Stochastic Differential Equations simulation
using GPU”, International Simulation Conference of India 2012.
WORK EXPERIENCE:
COMPUTATIONAL RESEARCH LABORATORIES LTD (CRL), PUNE
(April 2010 till date)
Designation: Engineer - Technology Innovation Software Group
Role Description: Development of methodology for performance optimization of various scientific applications

Image processing over GPGPU
Implemented the block co-relation algorithm to process SAR image generated from RADAR-SAT1 satellite
over GPGPU.
 Design of parallel version of given algorithm keeping SIMD platform in mind.
 Got ~20X speed-up compared to sequential version. Published and presented a paper at APSAR
2011 Korea.
 Tools/Platforms/Languages: Linux, Intel Dual Core, GPGPU – fx8800, C2070. C, CUDA-C.

Random Number Generator using CUDA
Implemented Mersenne-Twister real time pseudo-random number generator on the NVIDIA GT series
GPGPU, using CUDA.
 Generated a solution to avoid a space crunch over GPGPU.
 Implementation runs 4x faster as compare to NVIDIA's SDK random number generator, and
Published a paper at the GTC 2011. San Diego, CA.
 Tools/Platforms/Languages: C, Cuda-C, GPGPU – C2070, GTX275.

Computational Finance Algorithms
Ported stochastic differential equation (SDE)s using Euler and Milstein scheme using CUDA.
 Compute European options on GPGPU using C and CUDA, Implement random number generator.
 Got 330X speed-up compared to the sequential version, published a paper.

Accelerating Whether Forecast application using GPGPU
Ported RRTM, WSM3 and YSU modules of WRF from FORTRAN to CUDA-C.
 Development of porting methodology for the FORTRAN based scientific application to CUDA.
Redesign the flow of computation over the data points.
 Got 10x , 54X and 8X speed-up for overall RRTM, WSM3 and YSU respectively. Published a paper
on the acceleration results to GTC2012.
 Tools/Platforms/Languages: C, FORTRAN, Cuda-C, GPGPU – C2070, GTX275.
COMPUTATIONAL RESEARCH LABORATORIES LTD (CRL), PUNE
(June 2008 to March 2010)
Designation: Member of Technical Staff - Information Security Group
Role Description: Design and implementation of cryptanalysis technique s using parallel computing.

Bandwidth and Profile Reduction of Unsymmetric Square Sparse Matrix.
The work was part of Discrete Logarithm Problem (DLP) project.
 Implemented RCM algorithm, which uses permutation technique to find largest diameter of the
graph representation of the matrix.
 Solved 192 bit discrete log problem using CRL’s supercomputer ‘EKA’.
 Tools/Platforms/Languages: Linux, Intel Dual core, C, LibTomMath.

GSM Cryptanalysis
 A5/1 Encryption algorithm
Recovery of session key used for encryption of plaintext by A5/1 with information about few
frame of conversation. Implemented a Time Memory Data Trade-off (TMDTO) attack to
cryptanalyze A5/1.
 Crypt analysis of COMP-128
Recovery of private key used by the SIM card. Designed and implemented a correlation attack.
Attack exploits structural design flows of COMP128 i.e lack of diffusion. It’s a chosen challenge
attack, which requires smart card to query about 150,000 times to recover secrete key.

LFSR Based cipher-text-only reconstruction of stream ciphers.
Work involves recovery of feedback polynomials of LFSRs, recovery of algebraic normal form (ANF) of the
combining function and lastly the initial conditions of LFSRs from given only cipher-text.
 Responsible for initial condition recovery phase. Implemented classical Fast correlation attack.
 Tools/Platforms/Languages: Linux, Intel Dual core, C.
DEPARTMENT OF ELECTRICAL ENGINEERING, IIT BOMBAY
Designation: Research assistant & System Administrator
(July 2005 to May 2008)
Work Area: Teaching assistance for control and computing lab and Linux system administration
 Responsible for maintaining and upgrading all the lab systems as well servers for various software’s as
well as user accounts. ( Fedora or Ubuntu ).
 Responsible for guiding undergraduate as well as postgraduate students in their course projects as well
as labs.
ACADEMIC PROJECTS
M.Tech Thesis
 Modeling and Implementation of Cryptological Primitives @IITB
M.Tech Thesis. Principal Investigator: Prof. Madhav Desai.
Work involves, Modeling cryptographic ciphers, writing a C-code according to the constraints of AHIR,
implementing it in a hardware and optimizing performance of AHIR to get best “Throughput / Area” .
R & D Project
 Cryptanalysis of Symmetric Ciphers using Algebraic linearization. @IITB
July 2006 – December 2006, Principal Investigator: Prof. V.R.Sule.
Developed a system theoretic approach for algebraic linearization and cryptanalysis of stream
ciphers. Tested methodology on reduced version of A5/1 GSM cipher.
B.E. Project
 Auto tuning of PID parameters using Fuzzy Logic.
Implemented fuzzy auto tune procedure to adjust on-line parameters of a conventional PID
controller located in the forward loop of the process.
Course Work during M.Tech
Applied Linear Algebra, Matrix Computation, Finite Fields and applications, Design and analysis of
algorithms (audit-Stanford), Computer Network, Network Security, Information System
Management, Non-linear Dynamical Systems, Multivariable Control Systems, Optimal Control,
Introduction to Cryptography, System Design.
Organizational Skills and Achievements
 Expert in writing project proposals. Involved in end to end solution of the projects.
 Outstanding team contribution to CAOS , SC10 team @CRL.
 Outstanding team player in SAR project @CRL.
 Hostel Color award during year 2007-08.
 Mess Councilor of Hostel-12&13. Student's Council in IIT Bombay for year 2007-08. ( 1000 Students Annual Budget more than Rs. 2/- crore ).
 Coordinator, Mardi Gras, the biggest festival conducted by 3 hostels IIT Bombay (Jan 2007).
 Production team head, in a Performance Art Festival, IIT Bombay: winning best production award.
Hobbies : Swimming, Drawing Sketches, Trekking, Photography.
References : Will be made available on request.