Lev Kurilenko
Hardware Engineer at Microsoft
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolLev Kurilenko is a hardware engineer at Microsoft with three years of professional experience building FPGA and accelerator solutions for Project Brainwave and Catapult. He brings hands-on FPGA IP development experience from CERN and Lawrence Berkeley National Lab, where his high-speed IO core and RD53A emulator work were adopted by multiple national labs and research groups. Lev pairs digital layout and RISC processor design with applied ML experience—implementing neural network accelerators, gesture recognition with configurable CNNs, and contributing inference-focused improvements to the widely used DeepSpeed projects. His background spans product engineering at Lattice, teaching advanced digital courses, and cross-continental technical collaborations, reflecting a rare combination of academic rigor (MS EE, 3.82 GPA) and production-grade firmware/IP delivery.
3 years of coding experience
3 years of employment as a software developer
Master of Science, Electrical Engineering, 3.82, Master of Science, Electrical Engineering, 3.82 at University of WashingtonBachelor of Science, Electrical Engineering, 3.94, Bachelor of Science, Electrical Engineering, 3.94 at Washington State University Vancouver
Github Skills (21)
pytorch10
parallelization10
python10
machine-learning10
inference10
huggingface-transformers10
deepspeed10
deep-learning10
nlp10
text-generation9
pytest9
compression8
model-optimization8
tensorflow8
tensorrt8
Programming languages (1)
Github contributions (5)
deepspeedai/DeepSpeedExamples
Oct 2022 - Jan 2023
Example models using DeepSpeed
Role in this project:
ML Engineer
ML EngineerContributions:63 reviews, 78 commits, 84 PRs in 3 months
Contributions summary:Lev's commits primarily focus on refactoring and enhancing Hugging Face inference examples within the DeepSpeed ecosystem. They reorganized code structure, added new examples for tasks like text generation, translation, and fill-mask, and integrated deepspeed.init_inference for optimized model deployment. Their work included incorporating new models, and added device comprehension to support more devices for the inference pipeline.
deep-learningpytorchdeepspeed
deepspeedai/DeepSpeed
Oct 2022 - Jan 2023
DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
Role in this project:ML Engineer
ML EngineerContributions:1 release, 125 reviews, 63 commits in 3 months
Contributions summary:Lev contributed to the DeepSpeed library by modifying existing code and adding new test cases to support and improve inference capabilities. The user focused on enhancing the injection policy testing framework, specifically incorporating tests for models like T5 and Roberta. Their work involved updating code related to replacing modules and ensuring backwards compatibility, along with removing unnecessary contexts related to zero-inference, showcasing a focus on optimizing inference performance.
billion-parametersfinetuningtrainingmixture-of-expertszero
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.