Nikita Smetanin
Staff Research Engineer at Together AI
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolNikita Smetanin is a Staff Research Engineer in London with 10+ years building production-scale ML systems across Conversational NLP, Computer Vision and large distributed training pipelines. He has led and grown teams at Replika and contributed to research and deployment at Meta AI (FAIR), working on high-impact projects like DINOv2 and protein language models ESM-2/ESMFold. Comfortable moving between research and engineering, he has a track record of shipping retrieval and generative dialogue systems, optimizing inference performance by up to 4x, and scaling multimodal LLM work at Together AI. An active open-source contributor, he helped release ESM-2 model code and weights, fixing FSDP and CPU compatibility issues that improved real-world usability. Trained in computer science and mathematics, he combines deep technical rigor with hands-on product collaboration and infrastructure optimization.
10 years of coding experience
12 years of employment as a software developer
Master of Science (MS) Computer Science and Mathematics, Master of Science (MS) Computer Science and Mathematics at Ural State University named after A.M.GorkyBachelor's degree Computer Science and Engineering, Bachelor's degree Computer Science and Engineering at Ural State Technical University
Stackoverflow
3reputation
1kreached
0answers
2questions
Github Skills (17)
pytorch10
python10
machine-learning10
deep-learning10
nlp10
computational-biology10
continuous-deployment9
ml-deployment9
cluster-computing7
parallel-computing7
scientific-computing7
nginx6
documentation6
ruby-on-rails6
shopify-app-bridge6
Programming languages (3)
Github contributions (5)
facebookresearch/esm
Aug 2022 - Nov 2022
Evolutionary Scale Modeling (esm): Pretrained language models for proteins
Role in this project:
ML Engineer
ML EngineerContributions:4 releases, 7 reviews, 9 commits in 2 months
Contributions summary:Nikita contributed to the release of the ESM-2 protein language model, including source code and model weights. They fixed setup and versioning issues, updated example code, and resolved an issue related to fairscale usage in an FSDP-specific inference example. Further contributions focused on fixing a layer index in the FSDP example and updating the versioning. Additionally, the user addressed CPU mode compatibility and formatting issues, improving the usability and maintainability of the codebase.
scale-modelingproteinspretrained-language-modelsesmscale
nikitos9000/VideoMoji
Apr 2017 - Aug 2017
Contributions:24 commits in 3 months
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.