Kopylov Aleksei
TL;DR
Deep Learning Engineer proficient in various domains: Computer Vision, NLP, Recommendation Systems, Time Series, Speech.
Specialise in the end-to-end development of ML systems, following to best MLOps practices. Led teams up to 4 developers.
Bachelor of Computer Science ULSU
CONTACT
π +995-595-098-129
π Linkedin
π¬πͺ Tbilisi, Georgia
π£οΈΒ Russian, English
STACK
ML EXPERTISE
Computer Vision: Pose Estimation, Object Detection, Classification, Object Tracking, OCR, GAN, Segmentation, AR
Text: LLM, NER, Transformers, NLP
Tabular: Time Series Analysis, Recommendation Systems, Forecasting, Classification, Clustering
Sound: Speech to Text, Denoising
CAREER [ 4 years of experience ]
Head of Data Science at Gurtam [2022-2024]
Spearheading team growth; recruiting, developing, and mentoring ML/DS specialists; coordinating hardware infrastructure requirements; proactively identifying and leveraging new ML applications.
Handling Big Data pipelines - 4 million real objects with hundreds of features (petabytes of data that constantly changing).
Integrated LLM to streamline Q&A on multimodal and dynamically evolving documentation, reducing support department's workload.
Reduced client overwhelm through tailored recommendation systems, leading to a lesser load on the business development department.
Pioneered integration of MLOps, CI/CD, DWH, and Data Marts in the production environment that significantly reduced the time of task completions and made solutions robust.
Performed in-depth research and analysis of telematic data that enhanced data-driven business decisions.
Machine Learning Engineer at Zebrains [2020-2022]
Led the architectural design and strategic roadmap development for ML modules, guiding projects from concept through to production.
Led the development of an AR + ML application for virtual fitting of rings and bracelets (3D Hand Pose Estimation), showcased at several exhibition.
Led the development of a Speech Recognition system for video calls
Developed a service for extracting and processing complex multimodal data from documents (OCR, NLP, NER).
Engineered a solution for Detecting and Tracking Cars with OCR.
Introduced MLOps in production environment.
Developed a Python framework for synthetic image dataset generation, substantially enhancing ML model accuracy.
Developed a Data Mart for the management team, enhancing data analysis and reporting capabilities.
University Lecturer (Machine Learning) at ULSU [2021-2022]
Numerous of my pupils now excelling in the ML industry.
Developed and instructed a ML course, with focus on hands-on exercises and production approaches.
Managed ML internships; judged ML Hackathon.