I lead the object tracking and fine-grained track classification team at Nuro. We're building robots that deliver stuff to your home! My team uses machine learning for sequence modeling and computer vision to track objects, and understand the fine-grained intent of the agents (pedestrians, cyclists, vehicles) in the world.
Previously, I got my Masters from the Robotics Institute at CMU, conducting research on multi-agent reinforcement learning (helping large robot teams learn to collaborate and be robust to individual failures), advised by Prof. Katia Sycara. I have an undergraduate degree in Electrical Engineering from IIT Kanpur, India along with a minor in Artificial Intelligence. While doing my undergrad I co-founded Agilo Technologies (now Stempedia), where I developed an educational hardware prototyping platform for students. Best way to get in touch with me is email.
Highlights
[Dec 2022] I'm now officially managing the tracking team, in addition to the track classification project.
[Dec 2021] I'm now heading the track classification effort at Nuro. We are hiring!
[July 2019] Started working as a machine learning engineer at Nuro. Nuro builds these self-driving vehicles for local B2C deliveries.
[Jan-May 2019] Graduated with my MS in Robotics, and published a few papers:
Defended my MS thesis on learning transferable cooperative behavior in multi-agent teams [ report | code | slides | ICML paper]
Presented work on improving conversational AI using team regularization at AAMAS 2019 (acceptance 24%) in Montreal, Canada. [ paper | code]
Presented work on using reinforcement learning for improving AI-human teamwork in Space Fortress at the AAAI-19 Workshop on Reinforcement Learning in Games in Hawaii, USA. [ paper | code]
[Sept 2018] Released 2 first-author preprints on arXiv:
[May-July 2015] Developed a new voting based classifier for machine fault diagnosis at the IDEA Lab, IIT Kanpur [Paper URL]
Papers
Learning transferable cooperative behavior in multi-agent teams [ arXiv | code ] with Sumit Kumar and Katia Sycara
International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2020)
Community Regularization of Visually Grounded Dialog [ arXiv | code ] with Swaminathan Gurumurthy, Vasu Sharma, Michael Lewis, Katia Sycara
International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2019) - Oral (24% acceptance)
Also presented at: ICML Adaptive Learning Agents Workshop (2018) link and CVPR Visual Dialog Workshop (2018) link
Challenges of Context and Time in Reinforcement Learning: Introducing Space Fortress as a Benchmark [ arXiv | code ]
with Ryan Hope, Katia Sycara
2019 AAAI Reinforcement Learning in Games Workshop link
Better Safe than Sorry: Evidence Accumulation Allows for Safe Reinforcement Learning [ arXiv ] with Abhinau Kumar V, Kyle Dunovan, Erik Peterson, Timothy Verstynen, Katia Sycara
Generalization ability of majority vote point classifiers for motor fault diagnosis [ arXiv ] with Nishchal Verma 2016 IEEE International Conference on Industrial and Information Systems (ICIIS)