Zhonghao Wang

Experience

• NVIDIA · Cosmos Lab

  2026–Current

  Senior Research Scientist

  Cosmos model — Lead sim-to-real video transfer generation.

• Adobe · Firefly

  2024–2026

  Senior Applied Scientist

  Firefly Image 4 /5 — Led post training on preference alignment and diffusion acceleration.

• I interned at Google Research & Cloud AI (2022, 2023), Amazon AWS (2020) and IBM Research (2019). I obtained Ph.D. from University of Illinois at Urbana-Champaign and worked with Thomas Huang, Mark Hasegawa-Johnson, and Humphrey Shi. My research focuses on world foundation models for physical AI, with particular interests in agentic generation, multimodal learning, and diffusion model alignment and acceleration.

Selected Research

• MoGAN: Improving Motion Quality in Video Diffusion via Few-Step Motion Adversarial Post-Training

  ECCV 2026

• MapReduce LoRA: Advancing the Pareto Front in Multi-Preference Optimization for Generative Models

  CVPR 2026 · Highlight

• Rethinking Global Text Conditioning in Diffusion Transformers

  ICLR 2026

• TurboFill: Adapting Few-step Text-to-image Model for Fast Image Inpainting

  CVPR 2025

• UnLoc: A Unified Framework for Video Localization Tasks

  ICCV 2023

• Interpretable Visual Reasoning via Induced Symbolic Space

  ICCV 2021

• Rethinking Text Segmentation: A Novel Dataset and A Text-Specific Refinement Approach

  CVPR 2021

• Differential Treatment for Stuff and Things: A Simple Unsupervised Domain Adaptation Method for Semantic Segmentation

  CVPR 2020

All publications on Google Scholar ↗

Education

• University of Illinois at Urbana-Champaign

  Ph.D., Electrical and Computer Engineering, 2024
  M.S., Electrical and Computer Engineering, 2019
  B.S., Electrical and Computer Engineering, 2017