Amrita Mathuriya

About Amrita Mathuriya

We are hiring analog engineers and material modeling experts to work on beyond CMOS technology development. Please message me on Linkedin if interested!I have filed 100+ patents and co-authored multiple publications as the primary author in the top tier conferences. I have programed and improved performance of various supercomputers optimizing hard CS workloads such as Quantum Montecarlo, Cosmological models. I work on cutting edge research problems while embedded in product groups. My work lies at the intersection of hardware and software. At Intel, I was the technology lead of AI/ML projects in a HPC/ML ecosystem development group at Intel. I represented Intel in Big data center collaboration working with NERSC (DOE lab at LBNL), Cray Inc. and university collaborators to demonstrate extreme performance for deep learning workloads on Cori supercomputer which is ranked 8th in top 500 list Nov. I directed a cross-Intel effort of innovative architecture development for artificial intelligence. Reported the findings to Intel\'s CEO and highest management committee. 17 patent applications filed in various areas of computer architecture, programmability, algorithm mapping and circuit design. My recent work (as first author) scaled TensorFlow based Cosmology deep learning application to 8K nodes of Intel Xeon Phi nodes with 3.5 PF sustained training performance. With these performance enhancements, for the first time we predicted three cosmological parameters of the universe with unprecedented accuracy. The work published in Supercomputing 2018 (top conference in HPC).https://arxiv.org/pdf/18•••••28.pdfPerformance optimization and scaling of Quantum Monte Carlo: Demonstrated upto 4.5x speedups on recent Intel processors and IBM Blue Gene/Q with memory-footprints reduction of up-to 3.8x enabling new material research.The work published (as first author) in top conferences Supercomputing 2017 and IPDPS

Experience

1. Co-founder and CTO

   Stealth Startup

   Jan 2019 — Present · Portland, OR, US

   Work on Ferroelectrics, build beyond CMOS memory and logic.

Education

• Indian Institute of Technology, Roorkee

  B. Tech, Computer Science

  2003 — 2007

• Georgia Institute of Technology

  Masters in Computer Science, HPC, Computational Sciences and Engineering

  2007 — 2009

Skills

• Scientific Computing
• High Performance Computing
• Scalability
• Parallel Programming
• Software Development
• Matlab
• C
• Parallel Computing
• Algorithms
• C++
• Deep Learning
• Computer Architecture
• Distributed Systems
• Numerical Analysis
• Computer Science
• Java
• Mpi
• Perl
• Computational Geometry