Ziyang Xie
Ziyang Xie
AI researcher and founder working on 3D vision, simulation, and generative models. CS PhD student @ UCLA; MSCS @ UIUC; B.S. @ Fudan University. Co-founder & CTO (Manifolds-AI); author of WorldGen, MV-Map, S-NeRF.
Projects
Ideas brought to life. A look at the products, companies, and creative works built from the ground up.
WorldGen
Open-source framework to generate full 3D scenes from text or images in seconds with support for real-time exploration and loop-closure. Claimed wide adoption and strong GitHub traction.
MV-Map: Offboard HD-Map Generation with Multi-view Consistency
Paper and project proposing an offboard HD-map generation pipeline that aggregates multi-view predictions with an uncertainty network and voxel-NeRF to improve map consistency for autonomous driving.
S-NeRF: Neural Radiance Fields for Street Views
NeRF adaptation for street-view data enabling neural reconstruction and rendering tailored to urban street scenes; foundational work that later contributed to S-NeRF++.
Vid2Sim: Realistic and Interactive Simulation from Video for Urban Navigation
Simulation-from-video pipeline for urban navigation that reduces sim-to-real gap and enables interactive realistic simulation from real-world video.
Manifolds-AI (极泛流形) — Co-founder & CTO projects
Led R&D to build controllable 3D image/video generation and simulation pipelines combining graphics rendering with generative models; productized scene generation and simulation technologies.
Frozen transformers as visual encoders
Research showing frozen transformers in language models can act as effective visual encoder layers; ICLR 2024 (Spotlight).
In the Spotlight
A collection of articles, interviews, and viral content where achievements have been shared.
Recognitions
Formal validation from the top. A showcase of official awards, grants, and government acknowledgements.
ICLR 2024 Spotlight
2024Paper 'Frozen transformers in language models are effective visual encoder layers' selected as Spotlight at ICLR 2024.
Peer-reviewed publications (ICCV, ICLR, CVPR, TPAMI)
2023–2025Multiple top-tier publications including ICCV 2023 (MV-Map), ICLR 2023 (S-NeRF), ICLR 2024 (Spotlight), CVPR 2025 (Vid2Sim), and TPAMI 2025 contributions.
Open-source project traction
2025WorldGen and related codebases received notable community attention (GitHub stars, Hacker News visibility) as reported on the personal site/CV.
Work Experience
Key roles that defined the journey.
Founder / Role: Stealth Startup
Current (listed on LinkedIn)Founder working on 3D scene generation / simulation / AI productization in stealth.
Co-Founder & CTO
Sep 2023 – Oct 2025 (as listed on CV)Led R&D to build a controllable 3D image and video generation and simulation pipeline that integrates graphics and generative models; product and engineering leadership.
CS PhD Student (Research)
Current (UCLA)Doctoral research in 3D vision, simulation, and embodied AI; collaborations on Vid2Sim and related projects.
MS Student / Researcher
Sep 2022 – Jul 2023 (project periods listed on CV)MSCS student and researcher; worked with Prof. Yu-Xiong Wang on MV-Map and related 3D vision projects.
Autonomous Driving Research Intern
Jan 2022 – Jul 2022Internship focused on autonomous driving research and industry collaboration.
Undergraduate Research / Collaborations
2019 – 2023 (undergraduate period)Undergraduate research in computer vision and contributions to S-NeRF and related projects.
Education
Academic foundation and learning journey.
PhD Student, Computer Science
Doctoral studies in computer vision and embodied AI. Advisor: Prof. Bolei Zhou.
Master of Science in Computer Science (MSCS)
Graduate studies; research with Prof. Yu-Xiong Wang on MV-Map and 3D vision.
Bachelor of Science, Computer Science
Undergraduate degree; conducted research with Prof. Li Zhang.
Socials
Connect and explore.