dp

Kaustubh Sridhar

Hi there! I’m a research scientist at Google Deepmind.

I defended my PhD in Electrical and Systems Engineering at the University of Pennsylvania where I was advised by Insup Lee and closely collaborated with Dinesh Jayaraman.

I am interested in creating adaptive generalist agents that are parameter- and sample-efficient, for the digital and physical worlds. Towards this goal, I have worked on generative models and in-context learning, deep reinforcement and imitation learning (particularly from large offline datasets), and robust deep learning.

My recent work on a retrieval-augmented generalist agent and adding in-context adaptability to pre-trained VLAs directly aims for this goal.

For the physical world in particular, I believe that world models are the critical breakthrough necessary to transcend this goal and reach embodied AGI. For this reason, my efforts at Google Deepmind are focused on world modeling for robotics.

In the past, I have interned twice at AWS AI Labs and once at Ford and VW’s self-driving unit.

Before starting my PhD, I graduated with honors from the Indian Institute of Technology Bombay.

My CV can be found here.

Work Experience

Awards

  • Oral Presentation (top 1.8% of 11672 submissions) for REGENT at ICLR 2025
  • Best Paper Award Nomination for CODiT at ICCPS 2023
  • Top Reviewer (top 10%), NeurIPS 2022
  • Outstanding Reviewer (top 10%), ICML 2022
  • NSF Travel Grant, ICCPS, 2023
  • Student Travel Grant, American Control Conference, 2022
  • The Dean’s Fellowship, University of Pennsylvania, 2019
  • The Howard Broadwell Fellowship, University of Pennsylvania, 2019
  • SN Bose Scholarship, Gov. of India, 2018
  • KVPY Fellowship, Gov. of India, 2015
  • Dan 1 (black belt), Shotokan Karate Association, 2011

Selected Talks

  • I gave an Oral Talk at ICLR 2025 in Singapore on REGENT, our generalist embodied agent. Please find a recording of this talk (including Q&A) at this link.
  • I gave talks about training adaptive and sample-efficient generalist embodied agents to different audiences:
    • Apr 2025: NTU Singapore
    • Nov 2024: Google Deepmind
    • Nov 2024: Apple MLR

Media

Publications

Please select one of the following topics.