I am a Ph.D. candidate at The University of Washington (UW), Seattle where I do research in computer vision, machine learning, artificial intelligence, and natural language processing. More specifically, I am interested in designing efficient machine learning architectures for visual and textual data that can run on resource-constrained device (e.g., smartphones and embedded devices) with good generalization properties.
At UW, I am advised by Prof. Linda Shapiro and Prof. Hannaneh Hajishirzi. I am a part of the GRAIL lab, the UW NLP group, and the H2Lab. I also collaborate with The Tasker Center for Accesible Technology (TCAT), Allen Institute for AI (AI2), and XNOR.AI. Before coming to UW, I worked for about 5 years in Bangalore, India at Advanced Micro Devices (AMD) and Infosys Research Lab.
Real-time semantic segmentation using ESPNetv2 on iPhone7
Real-time object deteciton using ESPNetv2
- Jan 12, 2021: Our paper, DeLighT, is accepted at ICLR'21.
- Dec 7, 2020: Preprint of our paper, EVRNet: Efficient Video Restoration on Edge Devices, is now online.
- Nov 29, 2020: DiCENet accepted in Transactions of Pattern Analysis and Machine Intelligence (TPAMI).
- Oct 18, 2020: MedICat accepted in the Findings of EMNLP'20.
- Oct 18, 2020: Two papers accepted in ICPR'20.
- Aug 2, 2020: Preprint of our paper, DeLighT: Very Deep and Light-weight Transformers, is now online.
- Jun 15, 2020: Interning (remotely) at Facebook Reality Labs.
- July 25, 2020: Preprint of our paper, HATNet: An End-to-End Holistic Attention Network for Diagnosis of Breast Biopsy Images, is now online.
- Jan 29, 2020: Our paper, MLCD: A Unified Software Package for Cancer Diagnosis, has been accepted for publication at JCO Clinical Cancer Informatics.
- Dec 19, 2019: Our paper, DeFINE, has been accepted for publication at ICLR'20.
- Aug 10, 2019: Released the source code of our iOS app, ESPNetv2-COREML, that demonstrates the real-time performance of ESPNetv2 on edge devices.
- Aug 9, 2019: Our paper on breast cancer diagnosis is published in JAMA Network Open.
- Jun 8, 2019: Released the source code for our work on efficient network designs (ESPNetv2 and DiCENet). Check my Github page.
- Apr 30, 2019: Our paper for ASD classification got accepted at ICIP'19.
- Feb 24, 2019: Our paper, ESPNetv2, got accepted at CVPR'19.
- Aug 10, 2018: Our paper, PRU, got accepted at EMNLP'18.
- Jun 3, 2018: Our paper, ESPNet, got accepted for publication at ECCV'18.
- May 25, 2018: Our paper, Y-Net, got accepted at MICCAI'18.