CV

Summary

Machine Learning Engineer with over 7 years of experience in deep learning, machine learning, and data analytics, currently primarily focused on Generative AI in the Computer Vision area. I am greatly inspired by the open-source community and am seeking for a role which will unleash my passion for research and engineering.

Skills

Technical skills - Deep Learning, Computer Vision, GAN, Diffusion models, Natural Language Processing, Big Data Analysis, Research and Development

Experience

Prequel - Senior Computer Vision Research Engineer (03.2021 - Now)

• Implemented and deployed an img2img FCN GAN for facial modification and photo stylization. This architecture was heavily utilized to produce some viral high-load filters executing on our backend with almost 200+ RPS
• Optimized the proposed architecture for real-time on-device inference on iOS devices for video and camera processing (30+ fps on iPhone 11)
• Created an internal solution for generation of large-scale synthetic facial manipulation datasets through StyleGAN2/3 latent codes translation
• Participated in creation of the GIO photo app from the model’s side perspective. The app hit Top-50 Photo Apps in US App Store in a first week of its launch

Sberbank - Senior Data Scientist (08.2018 - 03.2021)

• Created an open-source fork of Yandex utility called tomita-parser for extraction of structured data from texts. My fork was written in Java and allowed us to run tomita-parser on top of Apache Spark and process millions of client’s transactions descriptions in minutes
• Created an internal library for Data Scientists and Data Analysts in my department to simplify the process of building data marts suitable for ML experiments
• Participated in quality improvement of “System for Prevention of Operational Defects”. Improved the approach for predicting the rotation angle for scanned documents, which helped us to reduce errors in subsequent classification model

Moscow State University - Software Engineer (09.2017 - 08.2018)

• Participated in the implementation of data storage system for EgyptSAT as a C# developer. I was responsible for developing a long-term storage service built on top of Microsoft Failover Clustering

Moscow State University - Machine Learning Engineer (05.2015 - 08.2018)

• Participated in RnD project in the field of behavioral biometrics, where I conducted several machine learning experiments in the area of anomaly detection and published a scientific paper as a co-author

Education

Moscow State University - Master’s degree in Intelligent Big Data Analysis

Thesis: Development of user recognition methods based on personal computer’s keyboard usage

Moscow State University - Bachelor’s degree in Applied mathematics and informatics

Thesis: Applying machine learning methods to user recognition using mobile device motion sensors

Additional courses

• Coursera, Functional Programming Principles in Scala (Martin Odersky, EPFL),
• MADE Big Data Academy, Certified Data Scientist, 1.5 years program

Projects

• 2023 - Open-Source Contributions - I’ve made several contributions to Hugging Face 🤗 peft library. These contributions were targeted at increasing peft compatibility with Stable Diffusion and LoRAs trained through other open-source scripts and were mentioned as significant community contributions in 0.4.0 release
• 2020 - MADE Big Data Academy graduation project - “Development of mobile OCR engine” was praised in the nomination “Best Product”. My team developed a Flutter mobile app for Android and iOS, which allows a user to search through the text extracted from gallery photos right on mobile device. I distilled a pretrained CRAFT model with EfficientNet backbone and trained my own CRNN network for Russian language in a seq2seq manner
• 2019 - Open-Source Contributions - I created a fork of Yandex tomita-parser which can work on top of Apache Spark framework. This allowed my company to solve tasks similar to Named Entity Recognition without putting lots of resources into data assessment
• 2018 - PicsArt AI Hackathon - I was a part of the team which took 4th place on an online stage (person segmentation) and 5th place on an offline stage (person beautification Telegram bot)