Scholar
Zijie Li
Google Scholar ID: ji7TXTMAAAAJ
Carnegie Mellon University
Citations & Impact
Citations
980
H-index
11
i10-index
11
Publications
16
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- Published several peer-reviewed papers such as 'Dual Diffusion for Unified Image Generation and Understanding', 'Scalable Transformer for PDE Surrogate Modeling', and more. Also involved in projects like applying diffusion models to high-fidelity flow field reconstruction.
Research Experience
- Spent a summer at ByteDance Seed Image-Generation, working on unified text/image generation with a single diffusion model.
Education
- PhD from the Mechanical and AI Lab at CMU, advised by Amir Barati Farimani.
Background
- Currently a researcher and engineer at ByteDance Seed Image-Generation. Interested in deep learning for physics simulation, from particle-based dynamics to Eulerian fluid simulation. Specifically, how to tweak popular neural network architectures like graph neural networks and Transformers using insights from numerical solvers to better suit specific simulation tasks.
Miscellany
- Serves as a journal reviewer for Nature Machine Intelligence, IEEE Transactions on Neural Networks and Learning Systems, etc., and as a conference reviewer for NeurIPS, ICLR, etc.
Co-authors