Awards and Fellowships
High-performance computing, Automatic tuning, Machine learning, Derivative-free approaches for numerical
optimization, and Stochastic local search algorithms for combinatorial optimization
- Auto tuning for sustained performance, energy, and resilience: Developed the SPAPT test suite, a
set of search problems for auto tuning. Conducted an experimental study of global and local search
algorithms and showed that local search algorithms are particularly attractive for large scale auto
tuning. Designed a trust-region search algorithm and showed its effectiveness. Proposed a parallel
machine learning approach for empirical performance modeling of scientific kernels and applications.
Currently, investigating the tradeoffs between power, energy, and runtimes of scientific codes under
novel multi- and many-core architectures such as Blue Gene Q, Intel®; Sandy Bridge, and Xeon PhiTM.
Developing multi objective optimization framework to optimize several conflicting code performance
- Extrapolating application characteristics on exascale architectures: Developed statistical learning
models that extrapolate application properties (e.g. instruction mix, memory size, memory
bandwidth) as a function of workload sizes. Derived a projected exascale workload that provides a
first estimate of a range of quantitative characteristics, including operation mix, compute and memory
intensity, and memory bandwidth. Applied the models to evaluate a new architectural approach,
processor under memory, that can potentially obtain 1.4- to 4.2-fold performance increases.
- Machine learning for high-performance computing applications: Developed an effective
machine-learning based load-balancing algorithm to find parameter settings for the Community Earth
System Model on IBM Blue Gene P and reduced the required core-hours usage by a factor of six.
Currently, developing an active learning methodology to remove the design point selection bias and
to identify the most interesting points for computationally expensive quantum chemistry simulations.
- Best paper award, 21st High Performance Computing Symposia (HPC), The Society for Modeling &
Simulation International, San Diego, April 2013.
- NSF/SIAM travel award, for SIAM Conference on Parallel Processing for Scientific Computing,
Savannah, GA, February 15-17, 2012.
- F.N.R.S. chargé de recherches fellowship, from the Belgian Funds for Scientific Research from
October 2010 to September 2013 (one of the most competitive postdoctoral fellowships in Belgium).
Declined in order to accept postdoc at Argonne National Laboratory.
- F.N.R.S. aspirant fellowship, from the Belgian Funds for Scientific Research, from October 2006 to
September 2008 (one of the most competitive Ph.D. fellowships in Belgium).
- Marie Curie fellowship, from the European Commission through the fifth framework human
resources and mobility program, from November 2004 to September 2006 (one of the most competitive
Ph.D. fellowships at the European level).
- International student scholarship, for the best academic performance from Otto-von-Guericke
Universität Magdeburg, Germany, from April 2003 to March 2004 (scholarship awarded every year to
the top two international students).
- First class with distinction, Bachelor of Engineering, Periyar University, India, May 2002.
CV as of Nov 2013. [PDF]
||Bachelor of Engineering in Computer Science
||Department of Computer Science
||Vinayaka Missions, Periyar University Salem, India