Member of Technical Staff

Hanqing Zhu 「朱汉卿」

| 📈 Experience | ⭐️ Publications | 🤟 Honors | 🐱 My Cats

I work on RL Science & Coding RL Bigrun at xAI. I am one of the key contributors for Grok 4.5 for its RL scaling and core algorithmic changes, one of our biggest leaps.

I earned my Ph.D. in ECE from UT Austin, advised by Prof. David Z. Pan and Prof. Ray T. Chen. I also worked closely with Prof. Zhangyang "Atlas" Wang on efficient AI.

Previously, I received my bachelor's degree from Shanghai Jiao Tong University with highest honor in 2020.

I was recognized as the ML and Systems Rising Star in 2025, received MLSys'25 Outstanding Paper Award (Honorable Mention), CVPR'25 AI4CC Workshop Best Paper Award, Texas ECE Achievement Award, and more honors.

I was raised, and completed all my schooling before college, in a small town in southwest China. I’m grateful for the opportunities that followed; the path I’m on now was beyond anything I imagined growing up. 🙂

My research aims to build efficient and scalable AI systems.

🤗 I develop more scalable algorithms for pretraining and RL by bridging foundational optimization theory
🤗 I push efficient AI and agent deployment

🚀 xAI | Jan '26 – Present | Member of Technical Staff

Team: Coding RL Bigrun & RL Science | Focus: Key contributor of Grok 4.5, Grok Build
♾️ Meta AI | May '25 – Dec '25 | Research Scientist Intern

Topic: Theory-driven Efficient Learning for RLVR | Advisor: Dr. Yuandong Tian, Dr. Zechun Liu, Dr. Kai Sheng Tai
♾️ Meta AI | May '24 – Oct '24 | Research Scientist Intern

Topic: Efficient Large-scale Pre-Training | Advisor: Dr. Jinwon Lee
💡 Lightelligence Inc. | May '23 – Sept '23 | Software Research Intern

Topic: Low-bit Chip-aware Training | Advisor: Dr. Weifeng Zhang
🧠 Google Brain | Jul '22 – Nov '22 | Student Researcher

Topic: RL-based Chip Placement | Advisor: Dr. Joe Jiang

I have published papers in top conferences in machine learning/ system/computer architecture/design automation, including MLSys, HPCA, NeurIPS, ICCV, COLM, DAC, ICCAD, and TCAD.

	ISO: Isospectral Optimization for RLVR Hanqing Zhu, Wenyan Cong, Zhizhou Sha, Sagnik Mukherjee, Xinyuan Song, Xiaoxia Wu, Yuandong Tian, Shiwei Liu, David Z. Pan, Zhangyang "Atlas" Wang Release soon
	The Path Not Taken: RLVR Provably Learns Off the Principals (Arxiv) Hanqing Zhu, Zhenyu Zhang, Hanxian Huang, DiJia Su, Zechun Liu, Jiawei Zhao, Igor Fedorov, Hamed Pirsiavash, Zhizhou Sha, Jinwon Lee, David Z. Pan, Zhangyang Wang^†, Yuandong Tian^†, Kai Sheng Tai^† Arxiv 2025. NeurIPS 2025 Workshop on Efficient Reasoning (Spotlight)* [ Paper / Blog / X post / 量子位 / 新智元 ] First theory-driven RLVR study and guidance for geometry-aligned RL optimization
	Can Test-Time Scaling Improve World Foundation Model? (Arxiv) Wenyan Cong^, Hanqing Zhu^, Peihao Wang, Bangya Liu, Dejia Xu, Kevin Wang, David Z. Pan, Yan Wang, Zhiwen Fan, Zhangyang Wang Conference on Language Modeling (COLM), 2025 [ Paper / Code ] First efficient test-time scaling for world foundation model
	APOLLO: SGD-like Memory, AdamW-level Performance (Arxiv) Hanqing Zhu^, Zhenyu Zhang^, Wenyan Cong, Xi Liu, Sem Park, Vikas Chandra, Bo Long, David Z. Pan, Zhangyang Wang, Jinwon Lee Conference on Machine Learning and Systems (MLSys), 2025 [ 🏆 Outstanding Paper Honorable Mention / Paper / Code / Hacker News / HuggingFace / LLaMA-Factory / FluxML / axolotl / 机器之心 ] Theory-driven scalable memory-efficient training with new-recording memory efficiency
	PACE: Pacing Operator Learning to Accurate Optical Field Simulation for Complicated Photonic Devices (Arxiv) Hanqing Zhu, Wenyan Cong, Guojin Chen, Shupeng Ning, Ray Chen, Jiaqi Gu, David Z. Pan Conference on Neural Information Processing Systems (NeurIPS), 2024 [ Paper / Code ] Theory-grounded efficient and fast operator model for scientific simulation
	Lightening-Transformer: A Dynamically-operated Optically-interconnected Photonic Transformer Accelerator (Arxiv) Hanqing Zhu, Jiaqi Gu, Hanrui Wang, Zixuan Jiang, Zhekai Zhang, Rongxin Tang, Chenghao Feng, Song Han, Ray T. Chen, David Z. Pan IEEE International Symposium on High Performance Computer Architecture (HPCA), 2024 (Acceptance Rate: 18.3%) [ Paper / Code ] Hardware-software Co-design; First photonic transformer accelerator

Best Paper Award , CVPR AI for Content Creation Workshop, 2025
Honorable Mention Outstanding Paper , MLSys, 2025
ML and Systems Rising Stars (38 out of 150+), ML Commons, 2025
DAC Ph.D. Forum, DAC 2025
ICLR Notable Reviewer, ICLR 2025
MLSys Student Travel Award, MLSys 2025
Texas ECE Graduate Achievement Award, UT Austin 2024
UT Graduate School Continuing Fellowship Nomination (1 of 2 nominees in ECE), UT Austin 2024
1st Place in IEEE/ACM MLCAD FPGA Macro-Placement Contest, MLCAD, 2023
MLSys Student Travel Award, MLSys 2023
Winner of Robert S. Hilbert Memorial Optical Design Competition, Synopsys, 2022
DAC Young Fellow, 2021
Shanghai Outstanding Graduate, 2020
Hongyi Scholarship, 2019
Outstanding Undergraduate Scholarship, 2019
Samsung Scholarship, 2018
Zhiyuan College Honors Scholarship, 2018
1st Prize, National Mathematical Contest in Modeling, Shanghai Division, 2018
Academic Excellence Scholarship, 2017-2019

I have two lovely cats :)

This template is a modification to Jon Barron's website and Rishab Khincha's website.