Undergraduate Student

Carnegie Mellon University

Pittsburgh, PA 15213

I am an undergraduate student pursuing a Bachelor of Science in Mathematics and Computer Science at Carnegie Mellon University (CMU), with an expected graduation in May 2026. My research interests span Embodied AI, Spatial Intelligence and Spatial Data Infrastrutre, with additional focus on Machine Learning for Healthcare.

I am currently conducting research in both the Computer Science Department and the Robotics Institute at CMU. Under the supervision of Prof. Srinivasan Seshan and Prof. Anthony Rowe, I work on large-scale spatial data infrastructure and spatial data understanding using multimodal large language models (MLLMs).

Previously, I was also fortunate to work with Prof. Jean Oh and Prof. Jeffrey Ichnowski on soft robots, with Prof. Matthew Johnson-Roberson and Dr. Weiming Zhi on 3D reconstruction and diffusion models, and with Prof. Adam Perer on machine learning systems for healthcare.

Besides research, I’m a big basketball enthusiast 🏀 and an avid fan of travel 🌏 and photography 📷!

I am actively seeking opportunities for internships and graduate school. Feel free to contact me at email.

news

Jun 18, 2026	Our paper SOFTMAP has been accepted to IROS 2026!
Nov 08, 2025	My personal website is released!

selected publications

arXiv
Flame3D: Zero-shot Compositional Reasoning of 3D Scenes with Agentic Language Models

Sagar Bharadwaj, Ziyong Ma, Anurag Ghosh, Srinivasan Seshan, and Anthony Rowe

arXiv preprint, 2026

Abs arXiv Bib Website

Flame3D answers compositional, multi-hop spatial queries about 3D scenes by chaining inferences over structured visual-textual scene memory with spatial, visual, and external tool calls.
@article{bharadwaj2026flame3d, title = {Flame3D: Zero-shot Compositional Reasoning of 3D Scenes with Agentic Language Models}, author = {Bharadwaj, Sagar and Ma, Ziyong and Ghosh, Anurag and Seshan, Srinivasan and Rowe, Anthony}, journal = {arXiv preprint}, year = {2026}, publisher = {arXiv,}, tags = {spatial-intelligence} }
IROS
SOFTMAP: Sim2Real Soft Robot Forward Modeling via Topological Mesh Alignment and Physics Prior

Ziyong Ma, Uksang Yoo, Jonathan Francis, Weiming Zhi, Jeffrey Ichnowski, and Jean Oh

In Proceedings of the 2026 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2026), 2026

Abs arXiv Bib Website

SOFTMAP is a sim-to-real learning framework for real-time 3D forward modeling of tendon-actuated soft finger manipulators using topological mesh alignment, simulation pretraining, residual correction, and actuation calibration.
@inproceedings{ma2026softmap, title = {SOFTMAP: Sim2Real Soft Robot Forward Modeling via Topological Mesh Alignment and Physics Prior}, author = {Ma, Ziyong and Yoo, Uksang and Francis, Jonathan and Zhi, Weiming and Ichnowski, Jeffrey and Oh, Jean}, booktitle = {Proceedings of the 2026 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2026)}, year = {2026}, publisher = {IEEE}, tags = {robotics} }

RA-L

DOSE3: Diffusion-based Unified Out-Of-Distribution Detection on SE(3) Trajectories

Hongzhe Cheng^*, Tianyou Zhang^*, Ziyong Ma^*, Tianyi Zhang, Matthew Johnson-Roberson, and Weiming Zhi

IEEE Robotics and Automation Letters (RA-L), 2025

Also Presented at ICRA 2026

Abs Bib Paper

@article{ma2024pose3,
  title = {DOSE3: Diffusion-based Unified Out-Of-Distribution Detection on SE(3) Trajectories},
  author = {Cheng, Hongzhe and Zhang, Tianyou and Ma, Ziyong and Zhang, Tianyi and Johnson-Roberson, Matthew and Zhi, Weiming},
  journal = {IEEE Robotics and Automation Letters (RA-L)},
  year = {2025},
  paper = {https://ieeexplore.ieee.org/document/11278083},
  note = {Also Presented at ICRA 2026},
  tags = {robotics}
}

ML4H

TempoQL: A Readable, Precise, and Portable Query System for Electronic Health Record Data

Ziyong Ma, Venkatesh Sivaraman, Richard D Boyce, and Adam Perer

In Proceedings of the Machine Learning for Healthcare (ML4H), 2025

Also Presented at TS4H @ NeurIPS 2025

Abs Bib Paper Code

@inproceedings{ma2025tempoql,
  title = {TempoQL: A Readable, Precise, and Portable Query System for Electronic Health Record Data},
  author = {Ma, Ziyong and Sivaraman, Venkatesh and Boyce, Richard D and Perer, Adam},
  booktitle = {Proceedings of the Machine Learning for Healthcare (ML4H)},
  year = {2025},
  paper = {https://proceedings.mlr.press/v297/ma26a.html},
  note = {Also Presented at TS4H @ NeurIPS 2025},
  tags = {healthcare}
}

NeurIPS

Building 3D Representations and Generating Motions From a Single Image via Video-Generation

Weiming Zhi, Ziyong Ma, Tianyi Zhang, and Matthew Johnson-Roberson

In Proceedings of the 39th Conference on Neural Information Processing Systems (NeurIPS 2025), 2025

Abs arXiv Bib

@inproceedings{ma2024single,
  title = {Building 3D Representations and Generating Motions From a Single Image via Video-Generation},
  author = {Zhi, Weiming and Ma, Ziyong and Zhang, Tianyi and Johnson-Roberson, Matthew},
  booktitle = {Proceedings of the 39th Conference on Neural Information Processing Systems (NeurIPS 2025)},
  year = {2025},
  publisher = {NeurIPS,},
  tags = {robotics}
}

NINeS
OpenFLAME: A Federated Spatial Naming Infrastructure

Sagar Bharadwaj, Ziyong Ma, Ivan Liang, Michael Farb, Anthony Rowe, and Srinivasan Seshan

In Proceedings of the New Ideas in Networked Systems (NINeS) Conference, 2026, 2026

Abs arXiv Bib Website

In this paper, we present the design and implementation of OpenFLAME stands for Open Federated Localization and Mapping Engine, a federated spatial naming system, or in other words, a federated mapping infrastructure.
@inproceedings{ma2025openflame, title = {OpenFLAME: A Federated Spatial Naming Infrastructure}, author = {Bharadwaj, Sagar and Ma, Ziyong and Liang, Ivan and Farb, Michael and Rowe, Anthony and Seshan, Srinivasan}, booktitle = {Proceedings of the New Ideas in Networked Systems (NINeS) Conference, 2026}, year = {2026}, publisher = nines, tags = {system} }