Jiankai Sun

Email: jksun [at] stanford [dot] edu

Google Scholar | GitHub | LinkedIn | Twitter

Current Research Interests

Hi there! I am a fourth-year Ph.D. candidate and Interdisciplinary Graduate Fellow at Stanford University, advised by Mac Schwager. I am also affiliated with the Stanford Center for AI Safety. I have spent time at Meta Superintelligence Lab and Microsoft Research.

Research Interests:

Generative AI and LLM: Efficient reasoning [ACM'25, NeurIPS'25] and thinking [SynapseRoute'25], LLM architecture design, including attention mechanisms [NeurIPS'25, ACL'25a] and mixture-of-experts (MoE) models [ICLR'26], reinforcement learning for LLMs [ACL'25b], applications to automated theorem proving [TMLR'24] and safe AI agents [ArXiv'25a].
Embodied AI and Robotics: Develop, understand, and interact with multimodal world models to enable reasoning, perception, planning, and control [CoRL'25a, RAL'25, IROS'25, IJRR'24, ICRA'22, ICRA'21]. Explore neural representations for robot perception [CoRL'25b, FastSplat'24, RAL'24]. Leverage reinforcement and imitation learning to enhance long-horizon planning [CoRL'25c, CoRL'23]. Applications span robotic manipulation [CoRL'25c, CoRL'23, ICRA'21], autonomous driving [ITSC'23, ITSC'22, CoRL'20a, CoRL'20b, ICRA'20, ICCV'19], and general-purpose everyday tasks [CoRL'22, RAL'20].

News

[Jan. 2026] KERN MoE accepted to ICLR 2026
[Jan. 2026] Glad to give a talk on KERN MoE at NVIDIA
[Dec. 2025] Glad to give a talk on KERN MoE at Cohere Labs
[Nov. 2025] Serving as Area Chair for HRI 2026
[Oct. 2025] Glad to give a talk at Ling Team, Ant Group: Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel
[Sep. 2025] Three papers accepted to NeurIPS 2025: SAS, ArchScale, and NopeRoomGS
[Aug. 2025] Three papers accepted to CoRL 2025: ARCH, SIREN, and ParticleFormer
[Jun. 2025] GRaD-Nav accepted to IROS 2025 — a research contribution on drones!
[Jun. 2025] Joining Meta FAIR in Menlo Park as a Research Scientist Intern. Scheduled to attend CVPR 2025 and RSS 2025
[May. 2025] DAPE V2 and SeqPO accepted to ACL 2025
[May. 2025] We are organizing the 1st Workshop on Category-Level Object Pose Estimation in the Wild (WCLOP) @ ICCV 2025. Call for Papers!
[May. 2025] MixUCB accepted to RLC 2025
[Mar. 2025] Reasoning with Foundation Models accepted to ACM Computing Surveys
[Feb. 2025] We are organizing the 2nd Embodied Intelligent Systems Meet Generative-AI Era (MEIS) Workshop @ CVPR 2025. Call for Papers!
[Dec. 2024] Got into NVIDIA Graduate Fellowship Finalist
[Sep. 2024] LocATe accepted to Nature Communications Engineering
[Aug. 2024] Foundation Models in Robotics accepted to IJRR
[Jun. 2024] Lyra accepted to TMLR
[Jun. 2024] We are organizing the Coop-Intelligence Workshop @ ECCV 2024. Call for Papers!
[May 2024] Awarded Stanford Interdisciplinary Graduate Fellowship (SIGF)
[May 2024] Open-X-Embodiment has been awarded the Best Paper Award at ICRA 2024
[Mar. 2024] Will give a talk at Meta. See you in Seattle!
[Jan. 2024] We are organizing the WIHR Workshop @ ICRA 2024. Call for Papers!
[Sep. 2023] Uncertainty-Aware Planning accepted to NeurIPS 2023
[Sep. 2023] MimicPlay accepted to CoRL 2023 as Oral Presentation, Best Paper/Best Student Paper Awards Finalist, and Best Systems Paper Award Finalist
[Sep. 2023] Large AI Models accepted to JBHI
[Jul. 2023] SSTA2 accepted to ITSC 2023
[Jun. 2023] NeRF-Loc accepted to RA-L
[Sep. 2022] BEHAVIOR-1K accepted to CoRL 2022 as Oral Presentation and Best Paper Nomination
[Sep. 2022] Awarded Stanford Min Zhu and Susan Xu Engineering Fellowship
[Jun. 2022] SSTA accepted to ITSC 2022
[May. 2022] Egocentric Human Trajectory Forecasting accepted to RA-L and IROS 2022
[Jan. 2022] PlaTe accepted to RA-L and ICRA 2022
[Feb. 2021] MAIRL accepted to RA-L and ICRA 2021
[Dec. 2020] HiABP accepted to AAAI 2021
[Oct. 2020] DeepDecision and NSPS accepted by CoRL 2020
[Jun. 2020] View Parsing Network (VPN) accepted by RA-L and IROS 2020
[Feb. 2020] SegVoxelNet and RED accepted by ICRA 2020
[May. 2019] CamNet accepted by ICCV 2019

Publications (Google Scholar)

(* and † indicate equal contribution or alphabetical order)

Experience

2026.02 -, Research Intern, NVIDIA Research, Santa Clara, CA
2025.06 - 2026.01, Research Scientist Intern, Meta Superintelligence Lab, Menlo Park, CA
2025.01 - 2025.05, Research Intern, Microsoft Research, Redmond, WA
2024.06 - 2024.09, Research Intern, Autodesk, Inc., San Francisco, CA
2022.05 - 2022.09, Research Intern, Baidu, Inc., Beijing
2021.06 - 2021.11, Research Intern, Tencent Holdings Ltd., Shenzhen

Talks

[2026/01] Talk @ NVIDIA, Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel
[2025/12] Talk @ Cohere Labs, Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel
[2025/10] Talk @ Ling Team, Ant Group, Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel
[2024/03] Talk @ Meta, Egocentric Human Trajectory Forecasting With a Wearable Camera and Multi-Modal Fusion
[2021/10] Talk @ Max Planck Institute for Intelligent Systems, Localization and recognition of human action in 3D using transformers

Selected Honors

AAAI Doctoral Consortium
Ant InTech Technology Scholarship
Meshy Fellowship
NVIDIA Graduate Fellowship Finalist
Honorable Mention for the HRI Pioneers
Stanford Interdisciplinary Graduate Fellowship (SIGF)
ICRA 2024 Best Paper Award
ICRA 2024 Best Student Paper Finalist, Best Manipulation Paper Finalist Awards
CoRL 2023 Best Paper/Best Student Paper Awards Finalist, and Best Systems Paper Award Finalist
CoRL 2022 Best Paper Nomination
The World Artificial Intelligence Conference (WAIC) Yunfun Award
Stanford Min Zhu and Susan Xu Engineering Fellowship
NeurIPS Top 10% of High-scoring Reviewer
National Scholarship (highest honor) × 3 years
Tang Lixin Scholarship
SenseTime Scholarship
UCLA CSST Scholarship
Outstanding Winner, MCM/ICM
Vilfredo Pareto Award, MCM/ICM

Professional Service

Technical Committees / Workshops / Tutorials:
1st Workshop on Category-Level Object Pose Estimation in the Wild (WCLOP) @ ICCV 2025
2nd Multi-Agent Embodied Intelligent Systems Meet Generative-AI Era: Opportunities, Challenges and Futures (MEIS) @ CVPR 2025 [Video]
1st Workshop on Wearable Intelligence for Healthcare Robotics (WIHR): from Brain Activity to Body Movements @ ICRA 2024
1st Workshop on Cooperative Intelligence for Embodied AI @ ECCV 2024
IEEE RAS TC on Multi-Robot Systems
IEEE RAS TC on Computer and Robot Vision
Journal Reviewer:
IEEE Transactions on Robotics (T-RO)
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
IEEE Transactions on Image Processing (TIP)
IEEE Transactions on Neural Networks and Learning Systems (TNNLS)
IEEE Robotics and Automation Letters (RA-L)
IEEE Journal of Biomedical and Health Informatics (JBHI)
Data-centric Machine Learning Research (DMLR)
Conference Reviewer:
Robotics: Science and Systems (RSS) 2024
IEEE Conference on Decision and Control (CDC) 2023
International Conference on Learning Representations (ICLR) 2025, 2024, 2022
Conference on Neural Information Processing Systems (NeurIPS) 2024, 2023, 2021, 2020 (Top 10% of high-scoring reviewer)
International Conference on Machine Learning (ICML) 2025, 2024, 2021 (Expert Reviewer)
IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2024, 2021
International Conference on Computer Vision (ICCV) 2025, 2023, 2021
European Conference on Computer Vision (ECCV) 2024
Conference on Robot Learning (CoRL) 2025
IEEE International Conference on Robotics and Automation (ICRA) 2023, 2021, 2020
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2025, 2022
International Joint Conference on Artificial Intelligence (IJCAI) 2023
International Symposium of Robotics Research (ISRR) 2024
International Conference on Artificial Intelligence and Statistics (AISTATS) 2025
Area Chair:
ACM/IEEE International Conference on Human-Robot Interaction (HRI) 2026
Session Chair:
IEEE International Conference on Intelligent Transportation Systems (ITSC) 2023
Industrial Community:
HKSTP IDEATION

Press

Emerging Cyber Attack Risks of Medical AI Agents was covered by: South China Morning Post.
Decoder-Hybrid-Decoder Architecture for Efficient Reasoning with Long Generation was covered by: Microsoft.

South China Morning Post ×

05/14/2025: DeepSeek's AI in hospitals is 'too fast, too soon', Chinese medical researchers warn
Microsoft ×

07/09/2025: Reasoning reimagined: Introducing Phi-4-mini-flash-reasoning