Марина

MARYNA BAKUN

[открыть контакты](см. выше в блоке «контактная информация»)

[открыть контакты](см. выше в блоке «контактная информация»)

[открыть контакты](см. выше в блоке «контактная информация»)
certificate of mate academy:
https://bit.ly/bakun-certificate
Skills:
• Python
• OOP
• Git / GitHub / Git LFS
• PyTorch
• Computer Vision (EfficientNet)
• NLP (Transformers/DistilBERT)
• Multimodal Learning
• Django / ORM / REST framework
• Flask
• FastAPI
• Tesseract OCR
• Linux (WLS, Ubuntu)
• Command Line (CLI)
• Virtual Environments
• Docker / Docker Compose
• HTML / CSS
• SQLite / PostgreSQL
• NumPy / Pandas
• Rest (basic)
• JavaScript (basic)
• Node.js (basic)

Languages:
• English: Intermediate (B1)
• Ukrainian: Native (C1)

junior python Developer | Ai/ml engineer
SUMMary:
Junior Python Developer | AI/ML Engineer with hands-on experience in Linux (WSL, Ubuntu) and AI-based systems. Mate Academy Certified Python Developer with a solid understanding of software engineering principles and the ability to quickly learn complex technologies. Experienced in building and deploying multimodal machine learning applications that combine Computer Vision and NLP. Skilled in working with command-line tools, environment setup, and running inference pipelines on Linux systems. Interested in edge AI, real-time systems, and integrating ML models into practical applications.
EDUCATion:
Mate Academy | Python Developer (Professional Certificate)
Jan 2025 — Nov 2025 (Phase 1: Professional Certification)
Nov 2025 — Present (Phase 2: Advanced AI/ML Specialization))

Description:
• Successfully completed an intensive professional program, building deep expertise through 527+ hands-on coding tasks.
• Mastered the Python ecosystem, including Advanced OOP, Data Structures, and Algorithms.
• Gained practical experience in building and deploying web applications and working with relational databases.
• Post-certification focus (Nov 2025 – Present): Dedicated to Machine Learning and Deep Learning, developing multimodal AI systems (PyTorch, CV, NLP).
my portfolio projects:
Edge-Ready Multimodal Object Analyzer | MemeSense:
https://github.com/MiKPo4eLiK/Edge-Ready-Multimodal-Object-Analyzer/tree/main
Description:
• Developed a multimodal AI system for high-accuracy meme classification using integrated visual and textual data.
• Designed a hybrid deep learning architecture combining EfficientNet (Computer Vision) and DistilBERT (NLP) in PyTorch.
• Implemented a custom multimodal fusion layer to integrate image and text embeddings into a unified prediction pipeline.
• Built an OCR pipeline using Tesseract for automatic text extraction from images.
• Developed a CLI tool for running inference directly from the Linux terminal.
• Deployed and configured the project in a Linux environment (WSL, Ubuntu), including environment setup and dependency management.
• Created a web interface using Flask for real-time interaction with the model.
• Managed large model files using Git LFS for efficient version control.
Stack:
Python, PyTorch, Transformers (HuggingFace), EfficientNet, DistilBERT, Flask, Tesseract OCR, Linux (WSL, Ubuntu), CLI, Git, Git LFS

MemeSense | Multimodal AI Classifier:
https://github.com/MiKPo4eLiK/MemeSense
Description:
• Developed a multimodal deep learning system that classifies memes with high accuracy by analyzing both visual content and embedded text.
• Engineered a hybrid architecture combining EfficientNet-B2 (for Computer Vision) and DistilBERT (for NLP) using PyTorch.
• Implemented a custom Feature Fusion Layer to merge embeddings from different data modalities into a single classification pipeline.
• Integrated Tesseract OCR and OpenCV for automated text extraction and image preprocessing.
• Built a user-friendly web interface with Flask for real-time model inference and prediction visualization.
• Managed large-scale model weights using Git LFS, ensuring efficient version control for deep learning assets.
Stack:
Python, PyTorch, Transformers (HuggingFace), OpenCV, Flask, Git LFS, Tesseract OCR.

IT-Company Task Manager | Full-Stack Web App:
https://github.com/MiKPo4eLiK/IT-company-task-manager
Description:
• Developed a comprehensive task management system for IT teams, featuring role-based access control and project tracking.
• Built a robust backend using the Django framework, implementing complex Model-View-Template (MVT) architecture.
• Designed and optimized a relational database (PostgreSQL) to manage relationships between employees, tasks, and project types.
• Implemented advanced features: custom search filters, secure employee authentication, and dynamic task assignment.
• Deployed the application to Render, managing environment variables and static file hosting for a production-ready environment.
• Focused on Clean Code and SOLID principles to ensure the scalability and maintainability of the codebase.
Stack:
Python, Django, PostgreSQL, HTML/CSS, Git, Render.

Airport API Service | RESTful API & Containerizati
https://github.com/MiKPo4eLiK/airport-API-servise
Description:
• Developed a scalable RESTful API for managing airport operations, flight scheduling, and ticket booking systems.
• Leveraged Django REST Framework (DRF) to implement robust API endpoints with JWT-based authentication and permission levels.
• Containerized the entire application stack using Docker and Docker Compose, ensuring consistent development and production environments.
• Integrated Swagger UI (drf-spectacular) for interactive API documentation and testing.
• Managed a PostgreSQL database with complex relations, optimized through Django ORM migrations.
• Implemented secure order processing and real-time flight data retrieval with automated validation logic.
Stack:
Python, Django REST Framework, Docker, Docker Compose, PostgreSQL, JWT, Swagger.