Abstract

I am a Research Scientist at Sora (now WorldSim/Robotics), focused on full-stack research. Previously, I worked in the RL, multimodal, and post-training orgs, co-training major models in each; most remain unreleased for non-technical reasons.

I joined OpenAI in Feb/2025 after PhD, where I trained and demoed ChatGPT-Voice on GPT5 livestream.
During my PhD, I founded and led TurningPoint AI Research, a team of 10 students and professors across multiple labs. We were the first to reproduce DeepSeek-R1’s “aha moment” in multimodal perception and reasoning.

Education

[01/2020 - 12/2024] Ph.D. in Computer Science, University of California, Los Angeles
[09/2015 - 08/2019] B.S. in Computer Science and Statistics, University of Michigan, Ann Arbor
[09/2013 - 07/2015] B.S. in Finance Honors Program (transferred), Shanghai University of Finance and Economics

Awards (Selected)

Outstanding Graduate Student (1 per department) - UCLA CS Department, 05/2022.

Outstanding Paper Award (1/8) - ICLR, 04/2021.

Award of Excellence - Microsoft Research Asia, 09/2019.

Highest Distinction Graduate - The University of Michigan, 08/2019.

Fung’s Excellence Scholarship, UC Berkeley Graduate Admission Committee, 03/2019.

Outstanding Intern Award, SenseTime, 01/2019.

James B. Angell Scholar, The University of Michigan, 2017-2019.

Shanghai Scholarship, 2014

Renmin Scholarship - First Prize, 2014

Publications and Manuscripts

Concepts or Skills? Rethinking Instruction Selection for Multi-modal Models

Andrew Bai, Justin Cui, Ruochen Wang, Cho-Jui Hsieh

Preprint, 2025

[Paper]

Don't Think Longer, Think Wisely: Optimizing Thinking Dynamics for Large Reasoning Models

Sohyun An, Ruochen Wang, Tianyi Zhou, Cho-Jui Hsieh

NeurIPS, 2025

[Paper] [Proceedings]

R1-Zero’s “Aha Moment” in Visual Reasoning on a 2B Non-SFT Model

Hengguang Zhou*, Xirui Li*, Ruochen Wang, Minhao Cheng, Tianyi Zhou, Cho-Jui Hsieh, (*Equal Contribution)

Preprint, 2025

[Paper] [Code]

TurningPoint AI

MOSSBench: Is Your Multimodal Language Model Oversensitive to Safe Queries?

Xirui Li*, Hengguang Zhou*, Ruochen Wang, Tianyi Zhou, Minhao Cheng, Cho-Jui Hsieh, (*Equal Contribution)

ICLR, 2025

[Paper] [Project Website] [Code]

TurningPoint AI

Large Language Models are Interpretable Learners

Ruochen Wang, Si Si, Felix Yu, Dorothea Wiesmann, Cho-Jui Hsieh, Inderjit Dhillon

ICLR, 2025

[Paper] [Code]

Google

The Crystal Ball Hypothesis in diffusion models: Anticipating object positions from initial noise

Yuanhao Ban, Ruochen Wang, Tianyi Zhou, Boqing Gong, Cho-Jui Hsieh, Minhao Cheng

ICLR, 2025

[Paper]

TurningPoint AI

Solving for X and Beyond: Can Large Language Models Solve Complex Math Problems with More-Than-Two Unknowns?

Kuei-Chun Kao, Ruochen Wang, Cho-Jui Hsieh

EMNLP, 2024

[Project Website]

Understanding the Impact of Negative Prompts: When and How Do They Take Effect?

Yuanhao Ban, Ruochen Wang, Tianyi Zhou, Minhao Cheng, Boqing Gong, Cho-Jui Hsieh

ECCV, 2024

[Paper]

TurningPoint AI

MuLan: Multimodal-LLM Agent for Progressive and Interactive Multi-Object Diffusion

Sen Li, Ruochen Wang, Cho-Jui Hsieh, Minhao Cheng, Tianyi Zhou

Arxiv, 2024

[Paper] [Project Website] [Code]

TurningPoint AI

DrAttack: Prompt Decomposition and Reconstruction Makes Powerful LLM Jailbreakers

Xirui Li, Ruochen Wang, Minhao Cheng, Tianyi Zhou, Cho-Jui Hsieh

EMNLP, 2024

[Paper] [Project Website] [Code]

TurningPoint AI

One Prompt is not Enough: Automated Construction of a Mixture-of-Expert Prompts

Ruochen Wang*, Sohyun An*, Minhao Cheng, Tianyi Zhou, Sung Ju Hwang, Cho-Jui Hsieh (*Equal Contribution)

ICML, 2024

[Paper] [Project Website] [Code]

TurningPoint AI

On Discrete Prompt Optimization for Diffusion Models

Ruochen Wang, Ting Liu, Cho-jui Hsieh, Boqing Gong

ICML, 2024

[Paper] [Project Website] [Code]

Google

Ameliorate Spurious Correlations in Dataset Condensation

Justin Cui, Ruochen Wang, Yuanhao Xiong, Cho-jui Hsieh

ICML, 2024

[Paper]

Scaling Up Dataset Distillation to ImageNet-1K with Constant Memory

Justin Cui, Ruochen Wang, Si Si, Cho-jui Hsieh

ICML, 2023

[Paper]

FedDM: Iterative Distribution Matching for Communication-Efficient Federated Learning

Yuanhao Xiong*, Ruochen Wang*, Minhao Cheng, Felix Yu, Cho-Jui Hsieh, (*Equal Contribution)

CVPR, 2023

[Paper]

DC-BENCH: Dataset Condensation Benchmark

Justin Cui, Ruochen Wang, Si Si, Cho-Jui Hsieh

NeurIPS, 2022

[Paper] [Code] [Project Website]

Efficient Non-Parametric Optimizer Search for Diverse Tasks

Ruochen Wang, Yuanhao Xiong, Minhao Cheng, Cho-Jui Hsieh

NeurIPS, 2022

[Paper] [Code]

Generalizing Few-Shot NAS with Gradient Matching

Shoukang Hu*, Ruochen Wang*, Lanqing Hong, Zhenguo Li, Cho-Jui Hsieh, Jiashi Feng, (*Equal Contribution)

ICLR, 2022

[Paper] [Code]

Learning to Schedule Learning Rate with Graph Neural Networks

Yuanhao Xiong, Li-Cheng Lan, Xiangning Chen, Ruochen Wang, Cho-Jui Hsieh

ICLR, 2022

[Paper]

RANK-NOSH: Efficient Predictor-Based NAS via Non-Uniform Successive Halving

Ruochen Wang, Xiangning Chen, Minhao Cheng, Xiaocheng Tang, Cho-Jui Hsieh

ICCV, 2021

[Paper] [Code]

Rethinking Architecture Selection in Differentiable NAS

Ruochen Wang, Minhao Cheng, Xiangning Chen, Xiaocheng Tang, Cho-Jui Hsieh

ICLR, 2021

[Paper] [Code]

Outstanding Paper Award (1/8)

DrNAS: Dirichlet Neural Architecture Search

Xiangning Chen*, Ruochen Wang*, Minhao Cheng*, Xiaocheng Tang, Cho-Jui Hsieh, (*Equal Contribution)

ICLR, 2021

[Paper] [Code]

Miscellaneous

Reading and Martial Arts

I have a bit of a chaotic reading habit, jumping between around 10 books at the same time. Topics include but are not limited to history, philosophy, business, science and engineering.

I practice Katori Shintō-ryū (main branch), trained directly by current Shihan Dai (師範代).

Top