Leo Boytsov
Senior Research Scientist at Amazon Web Services (AWS)
Pittsburgh, Pennsylvania, United States
Summary
🤩
Rockstar🎓
Top SchoolLeo Boytsov is a senior research scientist and machine learning engineer with 13 years of experience bridging academic rigor and production systems, currently contributing to AWS CodeWhisperer in Pittsburgh. He holds a PhD from Carnegie Mellon and a master's in Applied Math & Cybernetics from MSU, and his work spans (un)natural language processing, dependency parsing, boosted trees, and efficient nearest-neighbor search. Previously he led research-engineering efforts at BCAI and 3M Health Care and has deep software roots from roles at NCBI, Yandex, and Deutsche Bank. An active open-source contributor, he added NMSLIB support and build automation to the widely used ann-benchmarks project, improving benchmarks for approximate nearest neighbor libraries. Known for speaking both PyTorch and C++, he combines low-level systems know-how with state-of-the-art ML research to ship reliable, high-performance tooling. Colleagues rely on him for translating complex research into maintainable production code.
13 years of coding experience
22 years of employment as a software developer
Doctor of Philosophy (PhD), Computer Science, Doctor of Philosophy (PhD), Computer Science at Carnegie Mellon UniversityLomonosov Moscow State University
Stackoverflow
1reputation
0reached
1answer
0questions
Github Skills (17)
benchmark10
c-language10
python10
nearest-neighbors10
benchmarking10
cprogramming-language10
data-structure9
algorithm9
algorithms9
cmake9
devops9
data-structures9
initialization6
dynamic-linking6
docker6
Programming languages (8)
Github contributions (5)
erikbern/ann-benchmarks
Jun 2015 - Jul 2016
Benchmarks of approximate nearest neighbor libraries in Python
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:19 commits, 8 PRs, 70 comments in 1 year 1 month
Contributions summary:Leo significantly contributed to the `ann-benchmarks` project by adding support for the Non-Metric Space Library (NMSLIB), a core component for benchmarking approximate nearest neighbor algorithms. They implemented NMSLIB integration and defined parameters for various NMSLIB methods, including SW-graph and HNSW, which directly impacted the benchmarking capabilities. Furthermore, the user configured and updated the installation process by modifying the `install.sh` script and the `nmslib.sh` script to include necessary dependencies, download and build NMSLIB, and set up the correct environment.
pythonnearest-neighbornearestdockerbenchmark
fast-pack/PyFastPFor
Feb 2018 - Sep 2021
Python bindings for the fast integer compression library FastPFor.
Contributions:43 commits, 1 PR, 39 pushes in 3 years 7 months
integer-compressionpythonintegerpython-bindingscompression
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.