Xiangru (Edward) Jian

This account is verified. Learn more

@EdwardJian2

CS PhDing @UWCheritonCS. Prev @SFResearch and @ServiceNowRSRCH. Working on agents (data agent and CUA), multimodal learning, and data management. Open to Work.

Waterloo, Ontario edward-jianqaq.github.io Joined October 2022

828 Following 309 Followers 263 Posts

Latest posts 5

Pinned post

Xiangru (Edward) Jian @EdwardJian2 Apr 13

Thanks, @_akhaliq, for posting our latest work on AI for Manufacturing. We are happy to have your feedback and together push this front even more in the near future! (yes! it's only a start😆)

AK Verified account @_akhaliq Apr 13

FORGE

Fine-grained Multimodal Evaluation for Manufacturing Scenarios

paper: huggingface.co/papers/2604.07…

View quoted post on X

Xiangru (Edward) Jian @EdwardJian2 1d

Honored to be selected as a 2026 IEEE BigData Rising Star, one of 18 researchers recognized this year🙌. My research at the University of Waterloo explores AI agents and data systems. Grateful to my advisor, collaborators, and mentors for their support!

Want to know more about my research, please click: edward-jianqaq.github.io for more detail🚀

Xiangru (Edward) Jian @EdwardJian2 Oct 5

Excited to share X-Tree, which turns reusable agent experience into a structured training signal for more efficient generalization. Glad to have contributed, and huge credit to @TonyCheng990417 for pushing this work forward!

Sitao Cheng @TonyCheng990417 Oct 5

Agents learn from flat action sequences (SFT, RLVR), so reusable experience go unused.

We mine them into a tree without LLMs, and train on it as data, reward and context.

X-Tree: Tokenizing Reusable Experience for Efficient Agent Generalization 🧵
🔗 sitaocheng.github.io/xtree/

View quoted post on X

Xiangru (Edward) Jian @EdwardJian2 Jul 28

Thank @_akhaliq for sharing our work! We believe code/program makes life easier for most of CUA tasks!

AK Verified account @_akhaliq Jul 28

StateAct

Program State, before Pixels, for Long-Horizon Computer-Use Agents

paper: huggingface.co/papers/2607.22…

View quoted post on X

Xiangru (Edward) Jian @EdwardJian2 Jun 24

Happy to contribute to 3 of 4 projects here and be co-first author of 2 of them. This is a significant part of my effort on multimodal learning for the past 2 years. Would like to thank the whole team without whom none of these can be possible! Look forward to see the next move from this wonderful team. And please stay tuned for my next effort too 😆

Sai Rajeswar Verified account @RajeswarSai Jun 23

𝗛𝗮𝗹𝗳 𝗮 𝗺𝗶𝗹𝗹𝗶𝗼𝗻 𝗱𝗼𝘄𝗻𝗹𝗼𝗮𝗱𝘀. 🎉
Huge milestone for our Enterprise AI efforts: We are closing in on 500K+ downloads since last year, with 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲𝗢𝗽𝘀-𝗚𝘆𝗺 being our latest release.

What makes this meaningful is the impact around it: seeing researchers and contributors pick up the work, stress-test it and push it in directions we did not anticipate. a reminder that addressing real problems compounds over time.

Current snapshot on @huggingface 🤗 :
• 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲𝗢𝗽𝘀-𝗚𝘆𝗺 (ICML 2026) — 20K
• 𝗚𝗿𝗼𝘂𝗻𝗱𝗖𝗨𝗔 (ICLR 2026) — 396K
• 𝗨𝗜-𝗩𝗶𝘀𝗶𝗼𝗻 (ICML 2025) — 42K
• 𝗕𝗶𝗴𝗗𝗼𝗰𝘀 (ICLR 2025) — 43K

@shiva_malay,@PShravannayak, @TheJishnuNair, @aarashfeizi , @EdwardJian2, @KevinQHLin , @kxli_2000 , @joanrod_ai , @patricebechard , @tianyu_zh , @gspandana , @AbhayPuri98 , @sagardavasam, @siris_krishna, @ServiceNowRSRCH , @Mila_Quebec , @chrisjpal

View quoted post on X