I am thrilled to announce that starting Fall 2024, I will be joining Princeton University as a tenure-track Assistant Professor in the ECE department!
I am a Ph.D. candidate in the Electrical Engineering department at Stanford University where I am advised by Prof. Mark Horowitz and collaborate with Prof. Euan Ashley and Prof. Benedict Paten. I conduct research at the intersection of computer architecture and systems, and, computational genomics for clinical and comparative applications.
I have led the computational team for the world's fastest genome diagnosis technique. I worked as an intern with the Architecture Research Group at NVIDIA Research in the summer of 2022 and the Hardware engineering group at D. E. Shaw Research in the summer of 2018.
I am a 2023 Forbes 30 Under 30 honoree in the Science category, a recipient of the 2022 NVIDIA graduate fellowship and the 2017 Barratt and Oakley Family fellowship. I am a 2021 Cadence Women in Technology scholar, 2019 AnitaBorg Grace Hopper student scholar, and part of the 2021 CRA-WP Grad Cohort for Women.
Previously, I received a Dual Degree (B. Tech. and M. Tech.) in Electrical Engineering from the Indian Institute of Technology, Bombay in 2017 along with the Akshay Dhoke Memorial Award. At IIT Bombay, I worked on my Master's thesis in the High-Performance Computing lab advised by Prof. Sachin Patkar. I contributed to the Pratham project which is the first student satellite project of IIT Bombay launched by the Indian Space Research Organization (ISRO) on September 26, 2016. I have been fortunate to participate in a semester exchange program at the Cooper Union for the Advancement of Science and Art, New York
I am a trained classical dancer with a Master's in Dance in Bharat Natyam from the Art Society, Mumbai.
A Scalable GPU-based whole genome aligner
Pairwise Whole Genome Alignment (WGA) is a crucial first step to understanding evolution at the DNA sequence-level. Pairwise WGA of thousands of currently available species genomes could help make biological discoveries, however, computing them for even a fraction of the millions of possible pairs is prohibitive – WGA of a single pair of vertebrate genomes (human-mouse) takes 11 hours on a 96-core Amazon Web Services (AWS) instance (c5.24xlarge). SegAlign – a scalable, GPU-accelerated system for computing pairwise WGA. SegAlign is based on the standard seed-filter-extend heuristic, in which the filtering stage dominates the runtime (e.g. 98% for human-mouse WGA), and is accelerated using GPU(s). Using three vertebrate genome pairs, we show that SegAlign provides a speedup of up to 14x on an 8-GPU, 64-core AWS instance (p3.16xlarge) for WGA and nearly 2.3x reduction in dollar cost. SegAlign also allows parallelization over multiple GPU nodes and scales efficiently.
A more-accurate co-processor for whole genome alignments with high speedup
Whole genome alignment (WGA) is an indispensable tool in comparative genomics to study how different life forms have been shaped by evolution at the molecular level. Existing software whole genome aligners require several CPU weeks to compare a pair of mammalian genomes and still miss several biologically-meaningful, high-scoring alignment regions. These aligners are based on the seed-filter-and-extend paradigm with an ungapped filtering stage. Ungapped filtering is responsible for the low sensitivity of these aligners but is used because it is 200x faster than performing gapped alignment, using dynamic programming, in software. We show that replacing ungapped filtering with gapped filtering increases the number of matching base pairs in alignments by up to 3x. Our accelerator, Darwin-WGA, is the first hardware accelerator for whole genome alignment and accelerates the gapped filtering stage. Implemented on an FPGA, Darwin-WGA provides up to 24x improvement (performance/$) in WGA over iso-sensitive software. An ASIC implementation of the proposed architecture on TSMC 40nm technology achieves up to 10x performance/watt improvement on whole genome alignments over state-of-the-art software at higher sensitivity, and up to 1,500x performance/watt improvement compared to iso-sensitive software.