Maurizio Monge
Research Scientist at Meta
England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolMaurizio Monge is a Research Scientist with 19 years of experience blending research-grade mathematics and production software engineering, currently optimizing geometric structures such as bundle adjustment at Meta. He brings deep expertise in optimization, sparse linear algebra and CUDA-accelerated C++ extensions, notably contributing performance-critical solver work to Facebook Research's Theseus library. His background spans stochastic dynamical systems, p-adic mathematical research, and applied ML/NLP, reflecting a rare combination of theoretical depth and practical engineering. Previously a principal mathematician and academic lecturer, he guides robust, high-performance implementations from mathematical models to batched GPU kernels. Colleagues rely on him for fast algorithms and numerically stable solvers that scale to real-world vision problems.
19 years of coding experience
Doctor of Philosophy (Ph.D.) Mathematics, Doctor of Philosophy (Ph.D.) Mathematics at Scuola Normale SuperioreBS-MS in Mathematics Mathematics, BS-MS in Mathematics Mathematics at Università di Pisa
Italian, French, English
Github Skills (11)
cuda10
sparse-matrix10
pytorch10
sparse-array10
c-language10
cprogramming-language10
sparse-file10
linear-algebra9
back-end-development9
deep-learning8
deeplearning-ai8
Programming languages (5)
Github contributions (5)
facebookresearch/theseus
Dec 2021 - Oct 2022
A library for differentiable nonlinear optimization
Role in this project:
Back-end & Performance Engineer
Back-end & Performance EngineerContributions:98 reviews, 8 commits, 13 PRs in 10 months
Contributions summary:Maurizio contributed extensively to the CUDA-based sparse LU solver, implementing and optimizing core components for differentiable nonlinear optimization. Their work involved developing C++ extensions with CUDA, focusing on batched matrix operations and autograd functions. They also added support for damping and multiple solver contexts, indicating a focus on performance and robustness. The contributions show expertise in sparse matrix computations, CUDA programming, and integration with PyTorch.
pytorchroboticsdifferentiablebilevel-optimizationimplicit-differentiation
maurimo/kanimaji
Mar 2016 - Dec 2022
Contributions:23 pushes, 2 branches, 6 comments in 6 years 9 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.