PhD Student in Computer Science
I am a fourth-year PhD student at the University of Cambridge, advised by Prof. Nic Lane. I also work closely with Colin Raffel at the Vector Institute and the University of Toronto. Prior to that, I obtained my MPhil in Advanced Computer Science at Cambridge.
My research focuses on:
How can we build automated and principled data pipelines that make foundation model pretraining more efficient, safer, more controllable, and more agentic?
Pretraining data curation and mixing [ICLR'26: ADAPT], data selection and attribution [NeurIPS'24: CLUES], synthetic data curriculum [NeurIPS'25: Decomp].
How can we build benchmarks and evaluation pipelines that measure model capabilities, expose failure modes, and guide better data/model iteration?
Reasoning [NeurIPS'24: MR-BEN], tokenization [ICLR'26: TokSuite], scientific discovery [SDE], hardware generation [ASP-DAC'25: HDL].
How can models specialize, compose, and route efficiently during test time instead of being retrained as one monolithic system?
Mixture-of-Experts specialization [ICML'26: DPSL], model MoErging [ICLR'25 Workshop Proposal: MCDC], decentralized training [MLSys'25: Photon], cascaded inference and model routing [ICLR'26: Cascadia], parallel/decomposed reasoning [ICML Workshop: Parallel Reasoning]; [NeurIPS'25: Decomp], fair revenue sharing for model collaboration [NeurIPS'25 Workshop: Generalist Collaboration].
I am always glad to connect about research and industry opportunities where my background may be useful, and I am flexible about location.
Experience
On Kimi model pretraining data, knowledge-intensive tasks, and data rewriting.
With Alessandro Sordoni and The Froggy Team, on modular and synthetic data generation for reasoning tasks and coding agents.
With Colin Raffel, Guilherme Penedo, and Hynek Kydlíček, on large-scale pretraining data rebalancing based on FineWeb; and with Nicolas Papernot on multi-agent mechanistic design.
Member, working on mechanistic interpretability and automated alignment research.
With Minjie Wang, Mufei Li, and Zheng Zhang, on Deep Graph Library and GNNLens2.
Recognition
News
Publications
Gul Sena Altintas*, Malikeh Ehghaghi*, Brian Lester, Fengyuan Liu, Wanru Zhao, Marco Ciccone, Colin Raffel
ICML 2026, Oral
TokSuite is a collection of models and benchmarks designed to isolate and study the impact of tokenization on language model behavior across English, Chinese, Turkish, Italian, and Farsi languages, as well as STEM and mathematical text. It includes fourteen models that share the same architecture, training data, training budget, and initialization but differ only in their tokenizers, alongside a set of benchmarks that evaluate performance under real-world perturbations that affect tokenization.
Leyla Mirvakhabova*, Babak Ehteshami Bejnordi*, Gaurav Kumar, Hanxue Liang, Wanru Zhao, Paul Whatmough
ICML 2026
Wanru Zhao, Yihong Chen, Yuzhi Tang, Wentao Ma, Shengchao Hu, Shell Xu Hu, Alex Iacob, Abhinav Mehrotra, Nicholas Lane
ICLR 2026
Youhe Jiang*, Fangcheng Fu*, Wanru Zhao*, Stephan Rabanser, Jintao Zhang, Nicholas Lane, Binhang Yuan
ICLR 2026
Wanru Zhao, Lucas Caccia, Zhengyan Shi, Minseon Kim, Xingdi Yuan, Weijia Xu, Marc-Alexandre Cote, Alessandro Sordoni
NeurIPS 2025
Wanru Zhao, Hongxiang Fan, Shell Xu Hu, Wangchunshu Zhou, Bofan Chen, Nicholas Lane
NeurIPS 2024
Wanru Zhao, Yihong Chen, Royson Lee, Xinchi Qiu, Yan Gao, Hongxiang Fan, Nicholas Lane
ICLR 2024
Yaxin Du, Rui Ye, Fengting Yuchi, Wanru Zhao, Jingjing Qu, Yanfeng Wang, Siheng Chen
ACL 2025
Matthew Macfarlane*, Minseon Kim, Nebojsa Jojic, Weijia Xu, Lucas Caccia, Xingdi Yuan, Wanru Zhao*, Zhengyan Shi, Alessandro Sordoni
ICML 2025 AI for Math Workshop
Zhongshen Zeng, Yinhong Liu, Yingjia Wan, Jingyao Li, Pengguang Chen, Jianbo Dai, Yuxuan Yao, Rongwu Xu, Zehan Qi, Wanru Zhao, Linling Shen, Jianqiao Lu, Haochen Tan, Yukang Chen, Hao Zhang, Zhan Shi, Bailin Wang, Zhijiang Guo, Jiaya Jia
NeurIPS 2024
Shengchao Hu, Wanru Zhao, Weixiong Lin, Li Shen, Ya Zhang, Dacheng Tao
AAMAS 2025
Jiahao Gai, Hao Mark Chen, Zhican Wang, Hongyu Zhou, Wanru Zhao, Nicholas Lane, Hongxiang Fan
ASP-DAC 2025
Wanru Zhao, Vidit Khazanchi, Haodi Xing, Xuanli He, Qiongkai Xu, Nicholas Lane
ICLR Workshop on Secure and Trustworthy Large Language Models, 2024
Alan Chan, Rebecca Salganik, Alva Markelius, Chris Pang, Nitarshan Rajkumar, Dmitrii Krasheninnikov, Lauro Langosco, Zhonghao He, Yawen Duan, Micah Carroll, Michelle Lin, Alex Mayhew, Katherine Collins, Maryam Molamohammadi, John Burden, Wanru Zhao, Shalaleh Rismani, Konstantinos Voudouris, Umang Bhatt, Adrian Weller, David Krueger, Tegan Maharaj
FAccT 2023
Zhangde Song, Jieyu Lu, Yuanqi Du, Botao Yu, Thomas M. Pruyn, Yue Huang, Kehan Guo, Xiuzhe Luo, Yuanhao Qu, Yi Qu, Yinkai Wang, Haorui Wang, Jeff Guo, Jingru Gan, Parshin Shojaee, Di Luo, Andres M Bran, Gen Li, Qiyuan Zhao, Shao-Xiong Lennon Luo, Yuxuan Zhang, Xiang Zou, Wanru Zhao, Yifan F. Zhang, Wucheng Zhang, Shunan Zheng, Saiyang Zhang, Sartaaj Takrim Khan, Mahyar Rajabi-Kochi, Samantha Paradi-Maropakis, Tony Baltoiu, Fengyu Xie, Tianyang Chen, Kexin Huang, Weiliang Luo, Meijing Fang, Xin Yang, Lixue Cheng, Jiajun He, Soha Hassoun, Xiangliang Zhang, Wei Wang, Chandan K. Reddy, Chao Zhang, Zhiling Zheng, Mengdi Wang, Le Cong, Carla P. Gomes, Chang-Yu Hsieh, Aditya Nandy, Philippe Schwaller, Heather J. Kulik, Haojun Jia, Huan Sun, Seyed Mohamad Moosavi, Chenru Duan
arXiv preprint, 2025
* Equal contribution.
This list is not comprehensive. For the full publication list, please visit Google Scholar.
My name is pronounced similarly to "One Rule." My Chinese name is Wanru Zhao (趙婉如), derived from the classical verse 「有美一人，婉如清揚」, symbolizing a spirit that is vivid, bright, and pure. My English name is Renee, though here is a fun fact: during my time as a competitive programmer in Olympiads and online contests (e.g., CodeJam, Codeforces), I used Ryan as my handle to avoid being judged based on gender.
I love reading. The works of Friedrich Nietzsche, Albert Camus, and Max Weber have deeply influenced me. My favorite sci-fi writer is Ted Chiang. I love poetry as well, and view coding as a form of writing verse. My research is driven by a dual purpose: seeking intellectual satisfaction (thus adhering to internal standards of truth rather than hacking external metrics) and fulfilling a responsibility to contribute a unique, indispensable perspective to the world (thus ensuring that technology benefits the public and creates genuine social impact).
I enjoy traveling, mountaineering, and hiking. My journey has led me to reside in Cambridge, Montreal, Toronto, Chicago, Shanghai, and Beijing, and I eagerly look forward to embracing more diverse cultures. I am also an avid snow mountain enthusiast, and I climbed the impressive Peak Nochma of Minya Konka (5588m).
Beyond my research focus, I am interested in exploring the intersection of AI and the humanities in my spare time. Feel free to check out my past projects on computer music (which won a National Grand Prize) and AI for digital theater arts (Reimagine Copenhagen was exhibited at the Prague Quadrennial).
I am always open to new conversations and collaborations!