Rostyslav Geyyer
MTS Software Development Engineer at AMD
Houston, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolRostyslav Geyyer is a Machine Learning and Computer Vision software engineer with seven years of experience building high-performance ML runtimes and vision systems across industry and research. Currently an MTS Software Development Engineer at AMD, he focuses on performance-critical implementations for tensor operators and contributed examples and optimizations for grouped convolutions in the ROCm composable_kernel project, including low-precision data types like int4/int8 and bfloat16. Prior roles at Meta and Nabors reflect a blend of production ML engineering and applied computer vision, underpinned by research and teaching positions at Max Planck Institute and Otto-von-Guericke University. He holds advanced degrees in control engineering and completed research stays at MIT, giving him a strong foundation in mathematical modeling and systems thinking. Known for bridging research-grade algorithms with production performance, he often gravitates toward low-level optimization and mixed-precision workloads that materially speed up ML inference. Based in Houston, he brings a rare combination of academic rigor and hands-on kernel-level engineering to large-scale ML systems.
7 years of coding experience
8 years of employment as a software developer
Research stay, Control Engineering, Research stay, Control Engineering at Massachusetts Institute of TechnologyMaster of Science (MS), Electrical Engineering: Control Engineering, Master of Science (MS), Electrical Engineering: Control Engineering at Otto-von-Guericke-Universität MagdeburgBachelor of Science (BS), Systems Engineering, Bachelor of Science (BS), Systems Engineering at Donetsk National Technical University
English, German, Russian, Ukrainian, Spanish
Github Skills (11)
tensorrt10
neural-network10
cuda10
machine-learning10
c-language10
convolutional-neural-networks10
tensor10
tensorflow10
cprogramming-language10
performance-optimization10
operation10
Programming languages (3)
Github contributions (5)
ROCm/composable_kernel
Aug 2022 - Jan 2023
Composable Kernel: Performance Portable Programming Model for Machine Learning Tensor Operators
Role in this project:
ML Engineer
ML EngineerContributions:212 reviews, 68 commits, 86 PRs in 5 months
Contributions summary:Rostyslav focused on enhancing the `composable_kernel` repository by adding and optimizing examples for grouped convolution operations. They added new examples for `conv_fwd_bias_relu_add` across various data types, including int4, int8, bfloat16, float16, and float32. The user also addressed issues, refined existing code, and updated tests for batched gemm operations with int8 data.
kerneldeep-learningprogramming-modelmachine-learningtensor-learning
geyyer/FBGEMM
Feb 2022 - Jul 2022
FB (Facebook) + GEMM (General Matrix-Matrix Multiplication) - https://code.fb.com/ml-applications/fbgemm/
Contributions:40 pushes, 19 branches in 4 months
matrix-multiplicationlinear-algebrafacebookmultiplicationmatrix
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.