Wenbo (Vito) Zhu | OpusClip, UC Berkeley

01 / 08

About

Wenbo
(Vito) Zhu

Hello! I'm Wenbo (Vito) Zhu, currently the Founding AI Lead & Head of AI Research at OpusClip. I joined when the team had around 5 engineers, failed and pivoted, and now lead a 30+ cross-functional AI team delivering next-generation multimodal and generative video products. I initiated, led, and shipped several flagship systems, including:

OpusClip – No.1 AI Clipping Tool in the market
ClipAnything – First multi-modal clipping tool with prompting capabilities
Agent Opus – First video agent tailored for social media

Before OpusClip, I was a Senior ML Engineer at ByteDance/TikTok, where I served as a founding engineer of Gauthmath — building the world's first AI-based geometry solver with 100M+ downloads.

ICLR

On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification

Yongliang Wu*, Yizhou Zhou*, Zhou Ziheng, Yingzhe Peng, Xinyu Ye, Xinting Hu, Wenbo Zhu, Lu Qi, Ming-Hsuan Yang, Xu Yang

International Conference on Learning Representations (ICLR), 2026.

Integrated by ms-swift, trl, llama-factory.

PDF Code

CVPR

Adapting Point Cloud Analysis via Multimodal Bayesian Distribution Learning

Xingyu Zhu, Liang Yi, Shuo Wang, Wenbo Zhu, Yongliang Wu, Beier Zhu, Hanwang Zhang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026.

NeurIPS

KRIS-Bench: Benchmarking Next-Level Intelligent Image Editing Models

Yongliang Wu, Zonghui Li, Xinting Hu, Xinyu Ye, Xianfang Zeng, Gang Yu, Wenbo Zhu, Bernt Schiele, Ming-Hsuan Yang, Xu Yang

Advances in Neural Information Processing Systems (NeurIPS), 2025.

PDF Code

NeurIPS

DuSA: Fast and Accurate Dual-Stage Sparse Attention Mechanism Accelerating Both Training and Inference

Chong Wu, Jiawang Cao, Renjie Xu, Zhuoheng Ran, Maolin Che, Wenbo Zhu, Hong Yan

Advances in Neural Information Processing Systems (NeurIPS), 2025.

PDF

CVPR

Number it: Temporal Grounding Videos like Flipping Manga

Yongliang Wu*, Xinting Hu*, Yuyang Sun, Yizhou Zhou, Wenbo Zhu, Fengyun Rao, Bernt Schiele, Xu Yang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025.

Adopted by Andrew Zisserman's work.

PDF Code

AAAI

Unlearning Concepts in Diffusion Model via Concept Domain Correction and Concept Preserving Gradient

Yongliang Wu*, Shiji Zhou*, Mingzhuo Yang, Lianzhe Wang, Wenbo Zhu, Heng Chang, Xinting Hu, Xiao Zhou, Xu Yang

AAAI Conference on Artificial Intelligence (AAAI), 2025.

PDF Code

AAAI

Video Repurposing from User Generated Content: A Large-scale Dataset and Benchmark

Yongliang Wu*, Wenbo Zhu*, Jiawang Cao*, Yi Lu, Bozheng Li, Weiheng Chi, Zihan Qiu, Lirian Su, Haolin Zheng, Jay Wu, Xu Yang

AAAI Conference on Artificial Intelligence (AAAI), 2025.

PDF Code

CVPR Highlight

VEU-Bench: Towards Comprehensive Understanding of Video Editing

Bozheng Li, Yongliang Wu, Yi Lu, Jiashuo Yu, Licheng Tang, Jiawang Cao, Wenqing Zhu, Yuyang Sun, Jay Wu, Wenbo Zhu

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025.

PDF Code

ACL

RSVP: Reasoning Segmentation via Visual Prompting and Multi-modal Chain-of-Thought

Yi Lu*, Jiawang Cao*, Yongliang Wu*, Bozheng Li, Licheng Tang, Yangguang Ji, Chong Wu, Jay Wu, Wenbo Zhu

Annual Meeting of the Association for Computational Linguistics (ACL), 2025.

Compared and cited by SAM 3.

PDF

Research Code

Side Projects

On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification

Adapting Point Cloud Analysis via Multimodal Bayesian Distribution Learning

KRIS-Bench: Benchmarking Next-Level Intelligent Image Editing Models

DuSA: Fast and Accurate Dual-Stage Sparse Attention Mechanism Accelerating Both Training and Inference

Number it: Temporal Grounding Videos like Flipping Manga

Unlearning Concepts in Diffusion Model via Concept Domain Correction and Concept Preserving Gradient

Video Repurposing from User Generated Content: A Large-scale Dataset and Benchmark

VEU-Bench: Towards Comprehensive Understanding of Video Editing

RSVP: Reasoning Segmentation via Visual Prompting and Multi-modal Chain-of-Thought

Services

Conference Reviewers

Get in Touch