Biography

I am pursuing a Ph.D. in Electrical and Computer Engineering at North Carolina State University under the guidance of Dr. Aydin Aysu. I am a member of the hardware security lab of NCSU called HECTOR. My research mainly focuses on a combination of physical side-channel analysis of embedded systems, machine learning, and computer architecture. Currently, I am working towards securing Machine Learning IPs against power and electromagnetic -based side-channel attacks.

I have served as a co-reviewer to Dr. Aydin Aysu in various top-tier conferences like ICCAD, HOST, FPGA, ReConFig, DATE, etc. I have also been a teaching assistant for the course Cryptographic Engineering and Hardware Security offered at NC State University. I helped with the designing and grading of assignments on varied topics like Differential Power Analysis of cryptographic implementations, hardware design of post-quantum secure implementations (like ring-LWE), etc. I hold a certification by Riscure Inc. on Side-Channel and Fault Analysis of Embedded Systems.

Before joining the Ph.D. program, I was an ASIC Design Engineer in the GPU low-power team of NVIDIA Graphics, Bangalore. I contributed to numerous successful ASIC tapeouts during my stint at NVIDIA helping both in the design and verification phases.

Interests

  • Physical Side-Channel Attacks and Faults
  • Machine Learning
  • Computer Architecture

Education

  • PhD in Electrical and Computer Engineering, 2022 (expected)

    North Carolina State University

  • BE (Hons) in Electrical and Electronics, 2014

    Birla Institute of Technology and Science-Pilani

  • MSc (Hons) in Physics, 2014

    Birla Institute of Technology and Science-Pilani

Projects

Publications

Quickly discover relevant content by filtering publications.

BoMaNet: Boolean Masking of an Entire Neural Network

This paper proposes the hardware design and implementation of a neural network classifier resistant to power-based side-channel attacks using gate-level Boolean masking.

MaskedNet: The First Hardware Inference Engine Aiming Power Side-Channel Protection

This paper demonstrates successful Differential Power Analysis (DPA) attacks on an FPGA-based binarized neural network classifier to extract the weights and biases. It also proposes effective masking and hiding-based countermeasures to thwart such attacks.

Contact