Nikita Ermolenko
Machine Learning Engineer at JetBrains
United States
Summary
🤩
Rockstar🎓
Top SchoolNikita Ermolenko is a Machine Learning Engineer with 8 years of experience building production ML systems, robotics platforms, and developer tooling, currently developing LLM-driven code completion and smart IDE features at JetBrains AI. He has a strong track record of shipping high-impact improvements—refactoring an A/B testing pipeline to run ~28× faster and leading a re-ranking engine that boosted Python import suggestion top-1 precision by 20% for hundreds of thousands of developers. Nikita’s background spans robotics research and public-facing exhibits, from real-time perception for autonomous drones to dynamic motion planning for interactive robotic arms, reflecting a blend of safety-critical systems and product-focused delivery. He is an active open-source contributor to Kotlin Deep Learning (KotlinDL), adding ONNX/CUDA inference support and model preprocessing fixes that extend on-device and mobile deep learning use cases. Comfortable across Python, Kotlin/Java, C++, and embedded/ROS stacks, he combines systems-level rigor with a taste for elegant, user-facing technical experiences.
8 years of coding experience
8 years of employment as a software developer
Master's degree, Robotics and mechatronics, A, Master's degree, Robotics and mechatronics, A at Far Eastern Federal University
Russian, English
Github Skills (7)
kotlin10
deep-learning10
tensorflow10
onnx10
gpu10
cuda9
computer-vision9
Programming languages (4)
Github contributions (5)
Kotlin/kotlindl
Apr 2022 - Dec 2022
High-level Deep Learning Framework written in Kotlin and inspired by Keras
Role in this project:
ML Engineer
ML EngineerContributions:81 reviews, 33 commits, 50 PRs in 7 months
Contributions summary:Nikita made significant contributions to the `kotlin/kotlindl` repository, which is a deep learning framework. The commits focused on enhancing the OnnxInferenceModel, including adding model output type checks and providing support for inference with CUDA execution providers. They implemented functionality to handle different output types and ensured the framework's compatibility with various execution environments. Moreover, the user added support for converting tensor data to images and fixed preprocessing errors for models such as EfficientDet and SSD models.
gpudeep-learninghigh-levelmachine-learningkotlin
ermolenkodev/keras-salient-object-visualisation
Feb 2018 - Sep 2021
Keras implementation of nvidia paper 'Explaining How a Deep Neural Network Trained with End-to-End Learning Steers a Car'.
Contributions:9 commits, 1 PR, 7 pushes in 3 years 6 months
end-to-endnvidiadeep-learningdeep-neural-networkmachine-learning
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.