Sarvesh Patil

Hello, welcome to my half quirky, half stereotypical academic website! I’m a PhD student in Robotics advised by Prof. Max Simchowitz. Prior to that, for my 1st 2 years of PhD and MS in Robotics at CMU, I was advised by Prof. Oliver Kroemer and Prof. Zeynep Temel.

Reinforcement learningDexterous manipulationRepresentation learning

As a PhD student I’m trying to understand the principles behind reinforcement learning, variational inference, and representation learning and how each can be effectively applied to learn highly specialized fine-grained & dexterous manipulation skills across a wide range of application domains. My recent work, OGPO, studies sample-efficient full-policy finetuning of generative control policies.

Previously, as a Master’s student, I developed an array of 64 soft delta robots in an 8×8 grid to push the limits of dexterous manipulation with non-anthropomorphic soft robots. For the 1st 2 years of my PhD, I developed a sample-efficient multi-agent behavior cloning and reinforcement learning pipeline with zero-shot sim2real deployment for planar object manipulation using the soft delta robot array.

Prior to joining CMU I used to work in HyperWorks Imaging, a small startup in Bangalore, where I worked on making Pix2Pix, Enhanced Super Resolution GANs, and vision-based inference pipelines to save a few million dollars for Sila Nanotechnologies developing carbon-based high energy density batteries. Before that, I did my Bachelors in Engineering in Electronics and Telecommunication in VES Institute of Technology (Mumbai University) on a Tuition Fee Waiver Scheme scholarship.

Please feel free to connect with me if you’re interested in learning actually dexterous manipulation policies, highly sample-efficient reinforcement learning, representation learning for cross task and cross embodiment transfer of learned skills.

Recently

Jul 5, 2026	Presenting OGPO: Sample Efficient Full-Finetuning of Generative Control Policies at ICML 2026! ✈️📍Seoul, S. Korea 🇰🇷
May 22, 2026	Started my summer internship at Toyota Research Institute (TRI)🤖! 📍Boston, MA
Sep 2, 2025	Switched Labs! Started working with Prof. Max Simchowitz! 📍Pittsburgh, PA
May 19, 2025	Started Summer Internship at Bosch Center for AI🤖! 📍Pittsburgh, PA
Aug 15, 2024	Co-taught a group of 25 high school students basics of building robots and writing controllers for the Feiyue Summer School Program @CMU 🤖! 📍Pittsburgh, PA
Nov 6, 2023	Organizing a workshop on Learning for Soft Robots at CoRL 2023 ✈️ 📍Atlanta, GA
Oct 1, 2023	Helped organize my 1st workshop on Causality in Robotics at IROS 2023 ✈️ 📍Detroit, MI

Selected research

All publications

OGPO: Sample Efficient Full-Finetuning of Generative Control Policies

Sarvesh Patil, Mitsuhiko Nakamoto, Manan Agarwal, Shashwat Saxena, Jesse Zhang, Giri Anantharaman, Cleah Winston, Chaoyi Pan, Douglas Chen, Nai-Chieh Huang, Zeynep Temel, Oliver Kroemer, Sergey Levine, Abhishek Gupta, Hongkai Dai, Paarth Shah, and Max Simchowitz

In International Conference on Machine Learning (ICML), 2026

arXiv PDF Code Website

Cite this paper

@inproceedings{patil2026ogpo,
  title = {OGPO: Sample Efficient Full-Finetuning of Generative Control Policies},
  author = {Patil, Sarvesh and Nakamoto, Mitsuhiko and Agarwal, Manan and Saxena, Shashwat and Zhang, Jesse and Anantharaman, Giri and Winston, Cleah and Pan, Chaoyi and Chen, Douglas and Huang, Nai-Chieh and Temel, Zeynep and Kroemer, Oliver and Levine, Sergey and Gupta, Abhishek and Dai, Hongkai and Shah, Paarth and Simchowitz, Max},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2026},
}

Distributed Dexterous Manipulation with Spatially Conditioned Multi-Agent Transformers

Sarvesh Patil

arXiv preprint arXiv:2609.06930, 2026

arXiv PDF

Cite this paper

@article{Patil2025MADDM,
  title = {Distributed Dexterous Manipulation with Spatially Conditioned Multi-Agent Transformers},
  author = {Patil, Sarvesh},
  journal = {arXiv preprint arXiv:2609.06930},
  year = {2026},
}

From Fold to Function: Simulation-Driven Design of Origami Mechanisms

Tianhui Han, Shashwat Singh, Sarvesh Patil, and Zeynep Temel

In IEEE International Conference on Soft Robotics (RoboSoft), 2026

arXiv PDF

Cite this paper

@inproceedings{han2026fold,
  title = {From Fold to Function: Simulation-Driven Design of Origami Mechanisms},
  author = {Han, Tianhui and Singh, Shashwat and Patil, Sarvesh and Temel, Zeynep},
  booktitle = {IEEE International Conference on Soft Robotics (RoboSoft)},
  pages = {843--850},
  year = {2026},
}

CURATE: Automatic Curriculum Learning for Reinforcement Learning Agents through Competence-Based Curriculum Policy Search in Structured Task Spaces

Tabitha Edith Lee, Nan Rosemary Ke, Sarvesh Patil, Annya Dahmani, Eunice Yiu, Esra’a Saleh, Alison Gopnik, Oliver Kroemer, and Glen Berseth

Transactions on Machine Learning Research (TMLR), Aug 2026

Publication PDF Code Website

Cite this paper

@article{lee2026curate,
  title = {CURATE: Automatic Curriculum Learning for Reinforcement Learning Agents through Competence-Based Curriculum Policy Search in Structured Task Spaces},
  author = {Lee, Tabitha Edith and Ke, Nan Rosemary and Patil, Sarvesh and Dahmani, Annya and Yiu, Eunice and Saleh, Esra'a and Gopnik, Alison and Kroemer, Oliver and Berseth, Glen},
  journal = {Transactions on Machine Learning Research (TMLR)},
  year = {2026},
  month = aug,
}

Enhancing Dexterity in Robotic Manipulation via Hierarchical Contact Exploration

Xianyi Cheng, Sarvesh Patil, Zeynep Temel, Oliver Kroemer, and Matthew T. Mason

IEEE Robotics and Automation Letters, 2024

arXiv PDF

Cite this paper

@article{cheng2023enhancing,
  title = {Enhancing Dexterity in Robotic Manipulation via Hierarchical Contact Exploration},
  author = {Cheng, Xianyi and Patil, Sarvesh and Temel, Zeynep and Kroemer, Oliver and Mason, Matthew T.},
  journal = {IEEE Robotics and Automation Letters},
  year = {2024},
  volume = {9},
  number = {1},
  pages = {390--397},
}

60 citations · Google Scholar, Sep 13, 2026

Linear Delta Arrays for Compliant Dexterous Distributed Manipulation

Sarvesh Patil, Tony Tao, Tess Hellebrekers, Oliver Kroemer, and F. Zeynep Temel

In IEEE International Conference on Robotics and Automation (ICRA), 2023

arXiv Publication PDF Website

Cite this paper

@inproceedings{patil2022linear,
  title = {Linear Delta Arrays for Compliant Dexterous Distributed Manipulation},
  author = {Patil, Sarvesh and Tao, Tony and Hellebrekers, Tess and Kroemer, Oliver and Temel, F. Zeynep},
  booktitle = {IEEE International Conference on Robotics and Automation (ICRA)},
  year = {2023},
  pages = {10324--10330},
}

13 citations · Google Scholar, Sep 13, 2026

DeltaZ: An Accessible Compliant Delta Robot Manipulator for Research and Education

Sarvesh Patil, Samuel C. Alvares, Pragna Mannam, Oliver Kroemer, and F. Zeynep Temel

In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2022

arXiv PDF

Cite this paper

@inproceedings{Patil2022DeltaZAA,
  title = {DeltaZ: An Accessible Compliant Delta Robot Manipulator for Research and Education},
  author = {Patil, Sarvesh and Alvares, Samuel C. and Mannam, Pragna and Kroemer, Oliver and Temel, F. Zeynep},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  year = {2022},
}