StateM
An agent-native runtime for harness scaling, organizing execution around durable states, phase-local context, checked transitions, recoverable runbooks, and versioned procedural practices.
Biography
I am a researcher working on agents, efficient deep learning, data-centric AI, and large-scale model training.
My research has led to work such as StateM, InfoBatch, Dataset Growth, Info-Coevolution, and MERIT, with research collaborations across Meta, TikTok, Ali Tongyi Lab, and Tencent Hunyuan Lab.
Interests
- Agents
- Deep learning efficiency
- Data
- Optimizer
- Large-scale training
- Multimodal learning
Education
- PhD student in Computer ScienceNational University of Singapore
- MS in Computer ScienceUniversity of Southern California
- BSE in Computer ScienceUniversity of Michigan
- BSE in Electronic and Computer EngineeringShanghai Jiao Tong University
Recent News
- 2026 [Paper] DD-Ranking, our benchmark for fair evaluation of dataset distillation, was accepted for an oral presentation at NeurIPS 2026 ED Track.
- Aug 2026 [Research] StateM introduced harness scaling and achieved 95.3% raw accuracy on Terminal-Bench 2.1.
- 2026 [Release] The HY-WU technical report was released, focusing on test-time parameter adaptation.
- Dec 2025 [Position] Joined Meta Platforms as a Research Scientist.
- 2025 [Paper] Info-Coevolution was accepted to ICML 2025, introducing a framework for efficient data annotation.
- 2024 [Paper] Dataset Growth was accepted to ECCV 2024, introducing a method for maintaining a growing multimodal dataset.
- 2024 [Paper] InfoBatch was accepted for an oral presentation at ICLR 2024, focusing on training-time data efficiency.
Selected Awards
- NeurIPS ED Track Oral
DD-Ranking · 2026 · One of 15 oral papers
- ICLR Oral
InfoBatch · 2024
- AAAI Distinguished Paper Award
CowClip · 2023
- 200k SGD Research Funding
TikTok · Efficient data-model co-evolution · 2024-2025
- Golden Prize
SJTU-UMJI Summer Design Expo · 2019
Experience
Industry & research collaborations
Research Scientist · Meta Platforms
Research projects for modern recommendation architectures and AI agents/workflows for engineering production.
Research Engineer Intern · TikTok
Efficient data-model co-evolution for continually improving models and maintaining high-quality growing datasets with affordable cleaning and active annotation.
Research Collaboration · Ali Tongyi Lab
Research collaboration on data curation for data-centric AI, including research ideation, experiments, and paper writing. This collaboration led to first-author papers at ICLR 2024 Oral and ECCV 2024.
Software Engineer Intern · Google
Designed and implemented a Terraform-based system for automatically setting up VM-based distributed training environments on cloud infrastructure.
Projects
InfoBatch
Lossless training speed-up through unbiased dynamic data pruning, with practical savings across classification, segmentation, detection, vision pretraining, diffusion, and instruction tuning.
Info-Coevolution
An efficient active learning and data-model co-evolution framework for growing datasets and improving models under realistic compute and annotation budgets.
Dataset Growth
An online algorithm for data cleaning and selection that maintains growing datasets with awareness of cleanliness, diversity, and sampling value.
Recent Publications
Google Scholar
- Citations
- 441
- h-index
- 6
- i10-index
- 3
DD-Ranking: Rethinking the Evaluation of Dataset Distillation
Zekai Li, Xinhao Zhong, ..., Ziheng Qin, ..., Yang You, Kai Wang.
NeurIPS 2026 ED Track, OralStateM: Reaching 95.3% Raw Accuracy, or a $15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling
Ziheng Qin, Yaxin Lu, Zhangyang Atlas Wang, Kai Wang.
arXiv:2608.15089, 2026Info-Coevolution: An Efficient Framework for Data Model Coevolution
Ziheng Qin, Hailun Xu, Wei Chee Yew, Qi Jia, Yang Luo, Kanchan Sarkar, Danhui Guan, Kai Wang, Yang You.
ICML 2025InfoBatch: Lossless Training Speed Up by Unbiased Dynamic Data Pruning
Ziheng Qin, Kai Wang, Zangwei Zheng, Jianyang Gu, Xiangyu Peng, Zhaopan Xu, Daquan Zhou, Lei Shang, Baigui Sun, Xuansong Xie, Yang You.
ICLR 2024 OralDataset Growth
Ziheng Qin, Zhaopan Xu, Yukun Zhou, Zangwei Zheng, Zebang Cheng, Hao Tang, Lei Shang, Baigui Sun, Xiaojiang Peng, Radu Timofte, Hongxun Yao, Kai Wang, Yang You.
ECCV 2024Preventing Zero-shot Transfer Degradation in Continual Learning of Vision-language Models
Zangwei Zheng, Mingyuan Ma, Kai Wang, Ziheng Qin, Xiangyu Yue, Yang You.
ICCV 2023