Роман

Roman Shypka
[открыть контакты](см. выше в блоке «контактная информация»)
[открыть контакты](см. выше в блоке «контактная информация»)
Roman Shypka | LinkedIn
github:romanshypka11-maker
Data Science Engineer with over 6 monts of experience in building microservice-based AI platforms
and translating complex business requirements into intelligent AI agents and data-driven solutions
using Machine Learning algorithms and localized LLMs. Experienced with Python, LLM / Data
Pipelines, ML and Deep Learning algorithms, PostgreSQL, FAST API and Docker. Adept at bridging
the gap between deep technical execution and business-critical communication.
EXPERIENCE
Yakudza Cars - Junior Data Science Engineer (Project-based ) Jan 2026 — Present
◾ Architected and deployed a comprehensive, end-to-end microservice-based AI platform from scratch
to automate automotive market analytics and B2B lead generation
◾ Engineered a high-throughput, asynchronous web scraping system tailored for Auto.ria and Copart,
bypassing complex anti-bot protections to collect and structure a Big Data dataset of 230,000+ unique
listings.
◾ Designed and optimized a relational database schema using PostgreSQL,integrating an asynchronous
ORM layer(SQLAlchemy 2.0 & asyncpg)to minimize query latency for complex lookups
◾ Developed and fine-tuned a predictive gradient boosting model using CatBoost to accurately estimate
fair market car values based on historical trends.
◾ Integrated an RT-DETR deep learning architecture into the pipeline for automated vehicle damage
localization and detection from auction imagery
◾ Implemented an intelligent Text-to-SQL interface powered by local LLMs, enabling non-technical users
to generate real-time database queries via natural language.
◾ Developed a multimodal evaluation system that links predictive tabular models with computer vision
pipelines to analyze vehicle history, technical data, and auction photos, generating automated
comprehensive assessment reports.
◾ Built and optimized a robust backend infrastructure using FastAPI, implementing asynchronous routing
and Pydantic validation to handle high-frequency API requests and orchestrate microservice
communication.
◾ Collaborated directly with product owners and automotive sales managers to gather functional
requirements, translating business insights into technical specifications.
◾ Isolated and containerized all ecosystem services using Docker and Docker Compose, streamlining
multi-stage builds and reducing final image sizes for stable production deployment.
EDUCATION
Ivan Franko National University of Lviv Sep 2025 — Jun 2029
◾ Bachelor of Computer Science
SKILLS & OTHER
◾ Programming Languages: Python; SQL
◾ Machine Learning & Analytics:Gradient Boosting ,Scikit-learn, Feature Engineering, Pandas, NumPy
◾ Tools & Infrastructure: Docker, Docker Compose, FastAPI, Asyncio, Git / GitHub
◾ Languages: English (Upper-Intermediate / B2), Ukrainian (Native)
◾ Certifications: Machine Learning Specialization (DeepLearning.AI), Networking Basics (Cisco)
◾ Deep Learning & GenAI: PyTorch, RT-DETR, YOLO, OpenCV, Retrieval-Augmented Generation (RAG),
Prompt Engineering, Local LLM Deployment (Qwen, Ollama)