YuDeng
Beijing, China
Summary
🤩
RockstarResearch Scientist based in Beijing with eight years of experience, currently a senior researcher at Microsoft Research Asia focusing on 3D generation, neural rendering, and Embodied AI. Combines academic rigor with pragmatic engineering — exemplified by contributions to the widely-cited microsoft/Deep3DFaceReconstruction project (CVPRW 2019), where they ported NumPy pipelines to TensorFlow, added rendering, and improved preprocessing and demo tooling. Skilled at turning research models into reproducible demos and production-ready code, bridging the gap between prototype and deployable systems. Comfortable across the full stack of 3D ML workflows from data loading and preprocessing to differentiable rendering and model integration. Known for a practical, results-oriented approach that accelerates research impact and makes complex neural-rendering ideas usable in real applications.
9 years of coding experience
Github Skills (6)
tensorflow210
tensorflow10
python10
machine-learning9
computer-vision9
numpy8
Programming languages (4)
Github contributions (5)
microsoft/Deep3DFaceReconstruction
Mar 2019 - Jul 2022
Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set (CVPRW 2019)
Role in this project:
ML Engineer
ML EngineerContributions:89 commits, 5 PRs, 76 pushes in 3 years 3 months
Contributions summary:YuDeng primarily focused on transitioning the 3D face reconstruction process from NumPy to TensorFlow, adding rendering functionality, and updating image preprocessing methods. They modified the `face_decoder.py` file to incorporate TensorFlow operations for 3D face reconstruction and rendering. Further contributions include updates to demo scripts and data loading, indicating a focus on integrating and demonstrating the reconstruction model.
reconstructionsupervised-learningsupervised3d-face-reconstructionsingle-image
microsoft/DIF-Net
Mar 2021 - Jul 2022
Contributions:23 commits, 15 pushes, 8 comments in 1 year 4 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.