Stas Versilov
Director & Founder at Extra
Samara, Samara Oblast, Russia
Summary
🤩
Rockstar🎓
Top SchoolStas Versilov is a technology entrepreneur and engineering leader with 14 years of experience, currently founding and directing Extra from Samara, Russia. With a background in automation (specialist degree) he blends systems thinking and product leadership across long-running initiatives since 2008. He contributes as a backend and ML engineer to the Elixir-Nx project, improving tensor operations and PyTorch integration—work that demonstrates hands-on numerical and ML tooling expertise. His contributions include optimizing core tensor primitives and adding numeric robustness like NaN checks, and even an example MNIST neural network implementation. Comfortable shifting between strategic direction and low-level algorithmic detail, he brings both management experience and active open-source engineering to bear. Colleagues can expect a founder’s pragmatic focus on reliable, production-ready solutions informed by deep technical involvement.
14 years of coding experience
specialist, Automation, specialist, Automation at Samara State Technical University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (9)
tensorrt10
pytorch10
tensor10
tensorflow10
backend10
operation10
back-end-development10
elixir10
numerical-computing10
Programming languages (5)
Github contributions (5)
elixir-nx/nx
Feb 2021 - Sep 2021
Multi-dimensional arrays (tensors) and numerical definitions for Elixir
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:32 reviews, 18 commits, 27 PRs in 7 months
Contributions summary:Stas primarily contributed to the Elixir-Nx project, focusing on implementing and refining tensor operations, particularly within the context of PyTorch integration (Torchx backend). Their work involved defining and optimizing core tensor functionalities like `sum`, `slice`, `reshape`, and `dot product`, and converting between different types of tensors. They also worked on improving the numerical accuracy of the project, by implementing NaN checks for floating point types, and included an example MNIST neural network implementation.
pytorchmulti-dimensionalxlajitgpu
versilov/matrex
Apr 2018 - Jul 2020
A blazing fast matrix library for Elixir/Erlang with C implementation using CBLAS.
Contributions:488 commits, 13 PRs, 346 pushes in 2 years 3 months
erlangblazingelixirmatrix-libraryblazing-fast
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.