Назар

Hnatyshyn Nazar | Data Scientist
23 Sakharoza
[відкрити контакти](див. вище в блоці «контактна інформація») [відкрити контакти](див. вище в блоці «контактна інформація»)
Academic Street, Lviv

ABOUT ME
I’m a 4th-year Computer Science student at Lviv Polytechnic, majoring in Computational Intelligence
of Smart Systems. I’m passionate about artificial intelligence and currently developing my skills in data science. I enjoy
learning new things, exploring innovative technologies, and in my free time, I play volleyball and computer games.

TECHNICAL SKILLS ONLINE PROFILES
Languages → Experience: Python, SQL, C#/C++ github.com/MsNazarino
Tech Stack → Activily Used: VSCode, Jupyter Nootebook(Anaconda), Visual [відкрити контакти](див. вище в блоці «контактна інформація»)
Studio, PyCharm, MSSQL, MySQL kaggle.com/msnazarino
Familiar: Docker, Git, MongoDB
Business Modeling → Understanding of business logic, workflows, and system LANGUAGES
structure Ukrainian: Native
English: Pre-Intermediate

EDUCATION & CERTIFICATION SOFT SKILLS
Lviv Polytechnic National University 📍2022 – 2026 (expected) Effective communication
Teamwork
Bachelor’s Degree in Computer Science (Smart Systems and Computational
Time management
Intelligence)
Problem-solving
Department of Automated Control Systems
Cisco Certification
CCNAv7: Introduction to Networks (link)

WORK EXPERIENCE
Database Fundamentals (Lviv Polytechnic, 2023)
Completed academic course on relational databases using MSSQL.
Independently developed small projects with MySQLand MongoDB for personal assignments.
Lviv Polytechnic National University
Course project - Theories & Methods of Computational Intelligence | Feb – Jun 2025
Gained hands-on experience with regression, classification, and clustering techniques using scikit-learn,
pandas, numpy and another libs.
Built and trained perceptron and multilayer perceptron (MLP) models with TensorFlow
Developed and evaluated basic convolutional neural networks (CNN) for image recognition.
Studied Transformer architecture and its application in NLP.
Explored fundamentals of reinforcement learning (RL) using OpenAI Gym environments.
Preprocessed English text corpora: tokenization, stop-word removal, normalization.
Lviv Polytechnic – Information Systems Design
Engineered a information system tailored to business and user specifications, utilizing modeling tools such as
UML, DFD, IDEF0/IDEF3, AllFusion Process Modeler, and ERwin.

PROJECTS
1.The project task was to predict calories by creating and comparing regression models, classification, and ensemble
methods, including Linear Regression, Decision Tree, and Random Forest. Grid search was used to optimize the
hyperparameters of the models.
2.The goal of this project was to uncover hidden patterns in data through clustering, using PCA for efficient
dimensionality reduction and visualization. I developed models and compared the results of K-Means and DBSCAN
models before and after PCA.
3.Implemented a fuzzy logic control model in Python using NumPy, scikit-fuzzy and Matplotlib to simulate human-like
decision-making under uncertainty. Defined membership functions, linguistic variables, and Mamdani inference
rules to compute and visualize system outputs through defuzzification.
4.Developed a single-layer Perceptron model in Python to perform binary classification on linearly separable data.
Implemented using PyTorch, NumPy, Matplotlib, and scikit-learn for data generation, preprocessing, and
visualization. The model incorporated activation functions (ReLU, Sigmoid) to explore non-linear transformations and
demonstrated weight optimization, bias adjustment, and decision boundary formation during the training process.
5.Processed and analyzed text data using NLTK, datasets, and NumPy, performing tokenization, stop-word removal,
and word embeddings generation with GloVe, followed by cosine similarity calculations to evaluate semantic
relationships between words.
6.Developed an image classification system in Python utilizing PyTorch and torchvision with a transfer learning
approach. The project employed pre-trained (ResNet, VGG), where initial layers were frozen and a custom classifier
head was constructed. Training optimization was performed using Adam.