Roman Shapovalov

London, England, United Kingdom

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Roman Shapovalov is a backend-focused software engineer with 15 years of experience based in London, driven to create software that improves lives or advances global knowledge. He contributes to prominent open-source 3D and geometry projects—such as Point Cloud Library (PCL) and Facebook Research’s PyTorch3D—where he’s improved core numerical stability, implemented descriptors like Spin-Image, and enhanced algorithms like Perspective-n-Points and weighted Umeyama. Roman’s work emphasizes correctness and performance in geometric computing, from fixing voxel-grid refactors to refining SE(3) and rotation conversions. Trained at Lomonosov Moscow State University, he blends rigorous academic grounding with practical, low-level engineering chops. Less obvious: he gravitates toward subtle, high-impact fixes—documentation corrections and API refinements—that make advanced libraries more robust and usable for downstream researchers and products.

14 years of coding experience

Lomonosov Moscow State University

English, Russian

Stackoverflow

Stats

2,795reputation

1.2mreached

20answers

11questions

Badges

machine-learning

top-5%

computer-vision

top-5%

opencv

top-5%

Github Skills (33)

algorithm10

pcl10

algorithms10

pytorch10

c-language10

geometry10

numerical10

python10

point-cloud-library10

data-structure10

numerics10

numerical-methods10

3d10

stability10

data-structures10

Programming languages (4)

JavaC++MLIRPython

Github contributions (5)

facebookresearch/pytorch3d

Mar 2020 - Sep 2022

PyTorch3D is FAIR's library of reusable components for deep learning with 3D data

Role in this project:

Back-end Developer

Contributions:7 reviews, 29 commits, 10 comments in 2 years 7 months

Contributions summary:Roman primarily contributed to improving the PyTorch3D library, focusing on numerical stability, bug fixes, and code refactoring. They addressed issues related to string formatting, introduced weighted Umeyama implementations, and corrected a sign issue within the rotation conversion functions. Their work also included enhancing the efficiency of the Perspective-n-Points algorithm and expanding the API with the SE(3) log. These contributions indicate a focus on the underlying mathematics and functionality of the library.

pytorchdeep-learningcomputer-graphicscomputer-vision3d-data

PointCloudLibrary/pcl

May 2011 - Dec 2011

Point Cloud Library (PCL)

Role in this project:

Back-end Developer

Contributions:16 commits in 6 months

Contributions summary:Roman contributed to the core functionality of the Point Cloud Library (PCL), focusing on implementing and improving features related to point cloud processing. Their work included adding shortcuts for filters and features, enhancing the `makeShared()` method for `PointCloud` objects, and implementing the Spin-Image descriptor. Additionally, the user addressed issues in existing components, such as fixing voxel grid refactoring issues and correcting documentation.

cpppoint-cloud-libraryc-plus-pluscomputer-visionpcl-library

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.

Request Free Trial