I am a second-year PhD student at Mila and at the Robotics and Embodied AI Lab (REAL), advised by Liam Paull. My interests are diverse; predominantly along the intersections of robotics, computer vision, deep learning, and computer graphics. I spent the summer of 2019 with the amazing research group led by Sanja Fidler at NVIDIA, Toronto, where I led the development of Kaolin: a 3D deep learning library for PyTorch. Prior to joining my PhD in 2018, I pursused a Masters degree in computer science at IIIT Hyderabad, India, where I was advised by K. Madhava Krishna.
I am interested in the design of embodied agents that have a deep understanding of the rules of the world (geometry and physics). Modern AI techniques have endowed agents with an amazing set of perceptual capabilities. However, they fall well short of truly understanding the physical processes that govern the world around. My research attempts to bridge multiple fields (machine learning, computer vision, robotics, graphics, and physics) to construct better perceptual models of the world.
In particular, my research revolves around differentiable realizations of processes such as classical robotics, computer graphics, vision, and physics, with an application to gradient-based learning. Apart from this, I also devote a substantial amount of research effort towards solving applied problems such as perception for autonomous vehicles, generative 3D modeling, scene understanding, and robot mapping. I have open spots for interns with a strong background in my areas of interest. If that sounds like you (and you’d like to visit Mila), please fill out this form.
When I’m not doing research, I love to spend time writing technical blogs, tutorials, and open-source code.
|Dec 12, 2019||One co-authored paper accepted to WACV 2020.|
|Dec 12, 2019||One co-authored paper accepted to Robotics and Automation Letters (RAL).|
|Nov 12, 2019||Excited to launch Kaolin: A 3D Deep Learning library for PyTorch Whitepaper! GitHub repo|
|Oct 23, 2019||Excited to announce a new preprint: gradSLAM! Here’s the project page|
|Sep 28, 2019||Two papers accepted (IROS 2019, RAL 2019).|
|May 13, 2019||Deep learning research internship at NVIDIA|
|Apr 22, 2019||Two preprints out!|
|Jun 29, 2018||Two papers accepted to IROS 2018!|
|Apr 18, 2018||Paper on unsupervised visual odometry accepted to a CVPR 2018 Workshop|
|Apr 11, 2018||New preprint on unsupervised deep visual odmetry learning!|