About Ting Cao

📣 News

• 09/21/2026: We hosted the ICCL Self-Evolving Physical Intelligence Open Day in Wuxi. Prof. Ya-Qin Zhang opened the event and introduced Physical Self-Improvement (PSI). We presented our closed-loop framework for self-evolving embodied intelligence, including Zetta and Zeva.
• 09/21/2026: We released Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation, the first framework to learn physical causal priors for robot manipulation from unlabeled egocentric video.

👋 About Me

Dr. Ting Cao is a Professor at the Institute for AI Industry Research (AIR), Tsinghua University, and also the Founder of Z-Trans AI startup. Her current research centers on embodied intelligence, with a particular focus on self-evolving physical intelligence: building closed-loop systems in which embodied model and agents continually learn from interaction, improve their causal understanding, and translate that knowledge into more generalizable behavior. Her broader interests include Edge AI, AI training and inference systems, and novel AI accelerators.

Her work spans the full embodied intelligence stack, from learning and reasoning algorithms to efficient systems and hardware. It has been published at leading systems venues, including ISCA, ASPLOS, MobiCom, MobiSys, NSDI, OSDI, PLDI, EuroSys, SC, and PPoPP, and AI venues including ICLR, NeurIPS, ICML, and ICCV.

She has numerous awards, such as 2012 ACM Research highlights, 2012 IEEE Micro Top Picks, 2021 ACM SIGMOBILE Research highlights, PPoPP’24 Best Paper Award, MobiSys’21 Best Paper Award, NAS’14 Best Paper Award, ICCD’10 Best Paper Award, and MobiSys’26 Best Paper Runner-Up.

Dr. Cao’s research has also significantly advanced AI on edge devices, enabling complex deep neural networks and large language models to run directly on consumer devices such as smartphones and PCs while reducing cloud operational costs. Her innovations have been integrated into products used by millions, including Microsoft Office, Windows, Bing, and Huawei HarmonyOS.

🔬 Research Highlights

🤖 Self-Evolving Embodied Intelligence

• Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence
• Zeva: In-Context Causal Learning for Generalizable Embodied Manipulation
• Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation

⚙️ Selected Systems and AI Work

• Low-bit and table-lookup techniques for LLM inference
• Large-model accelerators built with commodity DRAM
• StreamMind for continuous video understanding
• V-Droid, a near-real-time GUI agent
• The Cloud for Science paradigm

🎓 Education and Work Experience

Before join Tsinghua University, Dr. Cao worked as a Principal Researcher and Research Manager at Microsoft Research, leading the Edge AI direction. Prior to that, she was with the Compiler and Programing Language lab at Huawei, and the State Key Lab of Computer Architecture at the Institute of Computing Technology, Chinese Academy of Sciences.

She received her PhD from the School of Computing at the Australian National University, where she was honoured to be supervised by Prof. Steve Blackburn and Prof. Kathryn McKinley. Her doctoral research focused on the Software and Hardware co-design for energy efficiency. Turing Award winner Prof. David Patterson wrote a perspective article “For Better or Worse, Benchmarks Shape a Field”, recommending her work on quantitative energy analysis for modern software and hardware for the ACM Research Highlight, and integrated the results into the classic textbook “Computer architecture: A Quatitative Approach”.

🤝 Community Serving

• Dr. Cao is serving as an Associate Editor for the journal IEEE Transactions on Computers
• She is now also serving as the Vice Chair for ASPLOS’27.
• She has also served on the program committees of conferences such as MobiSys, PLDI, OOPSLA, HPCA, VEE, ChinaSys, ISMM, et al. .