Alexander Panin
Head Of Mobile at SaluteDev
Tashkent, Uzbekistan
Summary
🤩
Rockstar🎓
Top SchoolAlexander Panin is an experienced mobile engineering leader with 11+ years building production-grade mobile and ML-infused systems, currently heading mobile at SaluteDev and SberDevices. His background spans team leadership at Yandex where he shaped SpeechKit and large-scale audio data collection tools, and he brings deep technical chops in ML, quantization, and distributed LLM inference from active contributions to projects like bitsandbytes, Petals and Hivemind. Comfortable bridging research and production, he has implemented machine translation and RL coursework components, optimized quantized matmuls, and added tensor-parallelism and CPU quantization to speed LLM inference. Based in Tashkent with an advanced mathematical modeling education, he pairs rigorous academic training with pragmatic engineering, and his GitHub tagline “building the hivemind” reflects a long-standing interest in decentralized and collaborative ML systems.
11 years of coding experience
9 years of employment as a software developer
Специалист Математическое обеспечение и администрирование информационных систем, Специалист Математическое обеспечение и администрирование информационных систем at Ulyanovsk State Universityкандидат физико-математических наук 05.13.18 Математическое моделирование численные методы и комплексы программ, кандидат физико-математических наук 05.13.18 Математическое моделирование численные методы и комплексы программ at Tolyatti State University
Английский
Github Skills (27)
markdown10
machine-translation10
pytorch10
back-end-development10
python10
sphinx10
word-embeddings10
machine-learning10
lm10
large-language-models10
reinforcement-learning10
numpy10
llm10
deep-learning10
markdown-it10
Programming languages (7)
Github contributions (5)
bigscience-workshop/petals
Jun 2022 - Jan 2023
🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading
Role in this project:
Back-end & MLOps Engineer
Back-end & MLOps EngineerContributions:208 reviews, 473 commits, 140 PRs in 7 months
Contributions summary:Alexander contributed to the development of memory-efficient modes, quantization, and tensor parallelism for running large language models (LLMs) using a BitTorrent-style approach within the Petals repository. Their work involved modifying core code, adding quantization features, and introducing tensor-parallelism capabilities. The contributions include code changes related to model architecture and adding CPU quantization scripts for the bloom model to enhance inference performance, and support for a client version of the models.
llama2finetuningbittorrentllamahuggingface-transformers
learning-at-home/hivemind
Mar 2020 - Jan 2023
Decentralized deep learning in PyTorch. Built to train models on thousands of volunteers across the world.
Role in this project:Technical Writer & Documentation Specialist
Technical Writer & Documentation SpecialistContributions:2 releases, 813 reviews, 713 commits in 2 years 10 months
Contributions summary:Alexander primarily contributed to the project by creating and updating documentation. They made minor fixes to the README file, initiated a Sphinx documentation quickstart, and configured Sphinx to support Markdown. The user also added sections, restructured content, and updated documentation-related configuration files, indicating a focus on improving project documentation. Overall, the user streamlined the documentation process and improved the project's user experience.
pytorchhiveminddhtasynciovolunteers
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.