Experience

Sumsub

• United Kingdom
• IT Services and IT Consulting
• 200 - 300 Employee

• Lead Computer Vision R&D Research Engineer

  • Mar 2023 - Present

  Berlin, Germany Fraud and Deepfake Detection: Revolutionized liveness and Deepfake detection, improved fraud detection accuracy, pioneered bias-free data methods for global compliance. Managed advanced GAN models. * Enhanced advanced liveness detection technology, cutting false positives by 30% without impacting conversion rates. * Led the development of a Deepfake Detection system, enhancing fraud detection accuracy by 35%. * Spearheaded efforts to eliminate biases in data, utilizing Stable… Show more Fraud and Deepfake Detection: Revolutionized liveness and Deepfake detection, improved fraud detection accuracy, pioneered bias-free data methods for global compliance. Managed advanced GAN models. * Enhanced advanced liveness detection technology, cutting false positives by 30% without impacting conversion rates. * Led the development of a Deepfake Detection system, enhancing fraud detection accuracy by 35%. * Spearheaded efforts to eliminate biases in data, utilizing Stable Diffusion and ControlNet for synthetic dataset creation. * Managed a comprehensive suite of cutting-edge Deepfake Generator models. * Orchestrated the development of an Age Estimation module, significantly boosting accuracy and aiding in global compliance and expansion efforts. Show less



ZERO10

• United States
• Technology, Information and Media
• 1 - 100 Employee

• Senior Computer Vision R&D Engineer

  • Jul 2022 - Apr 2023

  Limassol (Lemesos), Cyprus Real-time 3D Human Shape & Pose Estimation: Developed algorithms for depth, human segmentation, and shape/pose analysis. Designed and implemented data infrastructure to optimize model training. * Maintained and enhanced a broad range of real-time mobile Computer Vision models including Pose Estimation and Segmentation. * Achieved a substantial improvement in segmentation accuracy for hand-related classes. * Enhanced real-time performance, doubling frame rates for pose estimation… Show more Real-time 3D Human Shape & Pose Estimation: Developed algorithms for depth, human segmentation, and shape/pose analysis. Designed and implemented data infrastructure to optimize model training. * Maintained and enhanced a broad range of real-time mobile Computer Vision models including Pose Estimation and Segmentation. * Achieved a substantial improvement in segmentation accuracy for hand-related classes. * Enhanced real-time performance, doubling frame rates for pose estimation models. * Optimized data acquisition processes, leading to significant improvements in model performance metrics. * Revolutionized the data loading pipeline, tripling efficiency and aiding in efficient model training. Show less



ByteDance

• China
• Software Development
• 700 & Above Employee

• Senior Computer Vision Research Engineer

  • Aug 2021 - Jul 2022

  Singapore R&D Engineer, Video Deduplication: Transitioned video processing pipeline to a temporal-based approach. Implemented Vision Transformers for image/video embeddings. Enhanced deduplication pipeline with text and vision-based filters. * Implemented robust image and text frame-based filters, reducing content duplication significantly. * Developed innovative video-level embeddings using Visual Transformers, further decreasing video duplicates. * Pioneered AI-driven techniques for… Show more R&D Engineer, Video Deduplication: Transitioned video processing pipeline to a temporal-based approach. Implemented Vision Transformers for image/video embeddings. Enhanced deduplication pipeline with text and vision-based filters. * Implemented robust image and text frame-based filters, reducing content duplication significantly. * Developed innovative video-level embeddings using Visual Transformers, further decreasing video duplicates. * Pioneered AI-driven techniques for short-form video de-duplication in a high-paced environment. Show less



Picsart

• United States
• Software Development
• 700 & Above Employee

• Computer Vision Research Engineer

  • Mar 2020 - Mar 2022

  Deep Learning Research Engineer, Visual Filters: Specialized in beautification, including hair recoloring, video auto-adjustments, and photo enhancement. Built pipelines and models for diverse visual transformations. * Led a diverse research team in Image Generation and GAN Distillation, pushing the boundaries of Normalizing Flows. * Created a photo-realistic hair-recoloring tool, enhancing the capabilities of an existing StarGAN framework. * Optimized GANs for mobile deployment… Show more Deep Learning Research Engineer, Visual Filters: Specialized in beautification, including hair recoloring, video auto-adjustments, and photo enhancement. Built pipelines and models for diverse visual transformations. * Led a diverse research team in Image Generation and GAN Distillation, pushing the boundaries of Normalizing Flows. * Created a photo-realistic hair-recoloring tool, enhancing the capabilities of an existing StarGAN framework. * Optimized GANs for mobile deployment, achieving unprecedented speed increases and parameter efficiency. Show less



Higher School of Economics

• Russian Federation
• Higher Education
• 700 & Above Employee

• Teacher on AI Generative Models

  • Jan 2020 - Jan 2022

  Generative Models in Machine Learning Teacher: Conducted reading seminars for BA and MS students at HSE University, enriching their understanding of advanced ML techniques and applications Fields covered: GMM, Monte Carlo processes, GANs, VAEs, Normalizing Flows, Diffusion models



CERN

• Switzerland
• Research Services
• 700 & Above Employee

• Machine Learning Researcher

  • Oct 2018 - Mar 2020

  LAMBDA Lab Engineer, CERN Project: Developed ultra-fast particle identification models for the next Large Hadron Collider launch. Authored a paper on GANs for particle data generation, advancing research in high-energy physics.

• Deep Learning Researcher

  • Jul 2019 - Sep 2019

  Switzerland Summer Tech Student, Neutrino Event Reconstruction: Applied deep learning techniques to innovate and enhance the process of neutrino event reconstruction in a cutting-edge research project.