Polina

POLINA VLADYMYRSKA
SOFTWARE ENGINEERING STUDENT
[відкрити контакти](див. вище в блоці «контактна інформація»)

EDUCATION
National University of Kyiv-Mohyla Academy, Ukraine |BSc in Software Engineering | 2023 – 2027
University of Toronto, Canada | Exchange Student in Computer Science | 2025 – 2026
Coursework: Artificial Intelligence, Machine Learning, Probability and Statistics, Algorithms and Data
Structures, Mathematical Methods for Image Processing, Linear Algebra, Natural Language Processing
CGPA: 89.26/100

AWARDS & SCHOLARSHIPS
Temerty Foundation KMA Exchange Program Scholarship University of Toronto | 2025-2026
Full-ride merit-based scholarship for an exchange year
Bodnar Scholarship University of Toronto | 2026
Merit-based award for academic excellence and community leadership

PROJECTS
Architectural Style Classifier using CNN
● Trained and compared ResNet18, ResNet50, EfficientNet_B0, ConvNeXt on a 24-class dataset,
achieving 78% accuracy
● Evaluated model performance and addressed training challenges. Deployed an interactive demo
Stack: Python, PyTorch, torchvision, Google Colab, scikit-learn, FastAPI, Docker, Hugging Face Spaces
Demo: https://huggingface.co/spaces/aksrvl/architectural-style-classifier
GitHub: https://github.com/aksrvl/architectural-style-classifier

Active Assistive Vision Scene Understanding | In Progress
● Implemented monocular camera pose estimation using ORB feature matching and Essential
Matrix decomposition
● Developed ROI-based exploration planning and camera motion feedback. Extending the pipeline
for active scene exploration and visual question answering
Stack: Python, OpenCV, NumPy, pytest
GitHub: https://github.com/aksrvl/active-assistive-vision.git

Monocular 3D Perception and Trajectory Forecasting
● Built a monocular pipeline combining object detection, tracking and metric depth estimation to
reconstruct approximate 3D positions and visualize trajectories in bird’s-eye-view (BEV)
● Implemented Kalman filtering and constant velocity trajectory forecasting, evaluated depth
accuracy against Virtual KITTI 2 ground truth
Stack: Python, PyTorch, OpenCV, NumPy, YOLO, ByteTrack, Depth Anything V2, Virtual KITTI 2
Demo: https://aksrvl.github.io/DepthMotion/ | GitHub: https://github.com/aksrvl/DepthMotion.git

Bachelor’s Thesis – Reliability of Monocular Depth Estimation for Camera Pose
Refinement | In Progress
● Investigating when monocular depth estimates improve or degrade camera pose refinement in
SfM pipelines
● Developing an evaluation framework comparing pose refinement without depth priors, with
monocular depth priors and with ground truth depth

SKILLS
Programming Languages: Python, Java, C++
Computer Vision & ML: PyTorch, torchvision, OpenCV, CNNs
Tools & Technologies: pandas, NumPy, Matplotlib, scikit-learn, Git, Docker
Languages: Ukrainian (Native), English (Advanced)

SOCIAL INVOLVEMENT
Buddy – StudentMobility, NaUKMA Sep 2026 – present
Member of Outgoing Global Volunteer Team – AIESEC in Kyiv Mar – Jun 2025