Mingrui Zhang
Researcher at Citadel
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolMingrui Zhang is a London-based researcher with 8 years of experience specializing in physics-informed machine learning, differentiable physical simulation and high-performance computing. He holds a PhD from Imperial College London and has applied his research across industry roles at Taichi Graphics, Tencent, Man AHL and now Citadel, tackling 3D GenAI, quant ML and embodied AI. On GitHub he has contributed to the ICLR 2020 DiffTaichi ecosystem—upgrading and bug-fixing differentiable MPM simulators to modern Taichi—and integrated Taichi-based ray marching, volume rendering and hash encoders into a popular DreamFusion NeRF repo to accelerate 3D reconstruction. Mingrui blends deep theory with hands-on compiler and performance work, routinely adapting examples and packing bitfield optimizations to squeeze hardware efficiency. He excels at turning physics priors and automatic differentiation into production-ready ML systems that bridge simulation and 3D perception.
8 years of coding experience
3 years of employment as a software developer
Imperial College LondonBachelor's degree, Engineering, 5%, Bachelor's degree, Engineering, 5% at Zhejiang University
Chinese, English, Japanese
Github Skills (12)
volume-rendering10
machine-learning10
differentiable-programming10
serf10
python10
graphics8
gpu8
simulation7
simulations7
stable-diffusion7
robotics5
windows-gui4
Programming languages (7)
Github contributions (5)
taichi-dev/difftaichi
Oct 2021 - Jan 2023
10 differentiable physical simulators built with Taichi differentiable programming (DiffTaichi, ICLR 2020)
Role in this project:
ML Engineer
ML EngineerContributions:7 commits, 14 PRs, 12 pushes in 1 year 3 months
Contributions summary:Mingrui primarily focused on upgrading and adapting various example scripts within the `difftaichi` repository to the latest Taichi version (v0.8.1), which involves significant code changes. The user's contributions include updating numerous example files, indicating a focus on maintaining compatibility with the core differentiable programming library. This upgrade touches several aspects of the examples, including those related to differentiable MPM (Material Point Method) simulations, demonstrating a strong involvement with the core functionality of the library. The user also fixed bugs to improve functionality.
iclrroboticsdifferentiable-programmingdifferentiablesimulators
ashawkey/stable-dreamfusion
Feb 2023 - Mar 2023
Text-to-3D & Image-to-3D & Mesh Exportation with NeRF + Diffusion.
Role in this project:ML Engineer
ML EngineerContributions:1 review, 3 PRs, 6 comments in 27 days
Contributions summary:Mingrui's commits primarily focus on integrating Taichi, a high-performance parallel computing language, into the existing NeRF (Neural Radiance Fields) framework. This includes implementing Taichi-based ray marching, volume rendering, and a hash encoder for efficient feature extraction. They also contribute to the integration of Taichi for packing bitfields. The changes aim to accelerate the core components of the NeRF pipeline, showcasing optimization efforts for 3D reconstruction.
pytorchdeep-learningguinerfdiffusion
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.