Святослав

SVIATOSLAV KUZNETSOV
COMPUTER VISION ENGINEER

https://github.com/SvyatKtsov [открыть контакты](см. выше в блоке «контактная информация»)

[открыть контакты](см. выше в блоке «контактная информация»)
✆ Kyiv, Ukraine

WORK EXPERIENCE EDUCATION
SEPTEMBER-NOVERMBER 2024 SEP 2022-JUNE 2026 | BS COMPUTER
SCIENCE
AI ENGINEER (COMPUTER VISION),
BUMPY Taras Shevchenko National
University of Kyiv
Worked with data - cleaning, collecting datasets
(images) for training and evaluating object detection
models with PyTorch and TensorFlow (though mostly
CAREER OBJECTIVE
Pytorch) Passionate Computer Vision
Scripting with Python and in-built libraries Engineer. Ready to use my skills and
Got 80-90%+ accuracy after training and fine-tuning knowledge of Python, C++ and other
YOLO models to perform detection of sensitive
tools to solve real-world problems
in Computer Vision and Robotics
objects

HACKATHONS
November 2025 SKILLS
ARC 2025 Winner | BEST Kyiv Board Computer Vision
of European Students of Technology C
Redesigned underperforming 15° lubrication setup with Visual Odometry (ORB, SIFT, BF
Matcher, FLANN, Camera Pose
an optimized powder-spray system for internal pipe
estimation)
coating achieving uniform coverage while being 10-20x Python
cheaper than alternative solutions proposed by other teams Git
PyTorch - model training,
testing, optimization
PROJECTS YOLO
AUGUST 2025 OpenCV
Numpy, Pandas, Sklearn
DENOISING AUTOENCODER
Math
WITH PYTORCH
Collected data, added Gaussian + yellowish noise, labels LANGUAGES
for images
English - B2/C1 (Upper-
Trained an autoencoder (CNN-based) model
Intermediate/Advanced)
Optimized it using AOT Autograd - computational graph
was created before training, which sped up model Ukrainian - Native (C2)
training Russian - Advanced (C1)
MARCH-MAY 2024

CLUSTERING CUSTOMERS USING
K-MEANS (FROM SCRATCH)
Implemented clustering algorithm K-means from
scratch using Euclidean metrics as the distance
critera between data points
Developed Linear Regression and wrote Pearson
coefficient from scratch with an accuracy of
Lin.Reg. of 95%+
Created a GUI as the main interface