Oleksandr

Oleksandr Dmytrenko
Python Developer / Machine Learning Engineer

Würzburg, Germany, [відкрити контакти](див. вище в блоці «контактна інформація») [відкрити контакти](див. вище в блоці «контактна інформація»)

Summary

Junior Python Developer and Machine Learning Engineer with practical experience in computer vision and deep learning.
Built and trained end-to-end ML pipelines for object detection and image generation using PyTorch and YOLOv8, including
dataset preparation, model training, evaluation, and visualization. Strong foundation in Python, data analysis, and applied
machine learning through academic and personal projects.

Experience

Machine Learning / Python Developer (Personal Project) 03/2026

Traffic Object Detection API — Computer Vision System
• Built an end-to-end computer vision pipeline for traffic object detection using YOLOv8.
• Converted and prepared the VisDrone dataset, implemented dataset preprocessing and annotation conversion scripts.
• Trained and evaluated object detection models using PyTorch and Ultralytics YOLOv8.
• Developed a FastAPI REST service for real-time object detection from uploaded images.
• Implemented an inference pipeline returning bounding boxes, classes, confidence scores, and annotated prediction
images.
• Designed a modular project structure separating data processing, training, inference, and API layers.

Deep Learning Engineer (Personal Project) 04/2025 - 05/2025
Self-Attention GAN for Satellite Image Generation
• Implemented a conditional Self-Attention GAN (SAGAN) in PyTorch for generating satellite images of aviation
infrastructure.
• Built and curated a custom dataset (~400 images across 4 classes) and implemented automated annotation generation
scripts.
• Designed generator and discriminator architectures with self-attention layers and spectral normalization.
• Trained GAN models using Hinge Loss and mixed-precision training (AMP) to improve stability.
• Evaluated model performance using FID and LPIPS metrics for perceptual image quality assessment.
• Visualized training dynamics, generated samples, and attention heatmaps for model interpretability.

Python Developer (Personal Project) 11/2025
AI Dataset Inspector — Python CLI Tool
• Developed a command-line tool in Python to analyze and validate image datasets.
• Implemented automatic detection of corrupted, empty, and unsupported image files.
• Built duplicate detection using SHA-256 hashing.
• Implemented automated unit tests using pytest.

Education

Bachelor Degree, Automation and computer-integrated technologies, 2021 - 2025
Kyiv Aviation Institute,

Exchange Student, Mathematics, 2022 - 2023
Julius Maximilians University Würzburg,
DAAD Scholarship
Skills

Python Machine Learning
Deep Learning Computer Vision
PyTorch YOLOv8
Data Analysis REST APIs
FastAPI Generative Adversarial Networks (GANs)
Git Dataset Preparation
NumPy PostgreSQL

Languages

English - Upper-Intermediate (B2) German - Intermediate (B1+)

Ukrainian - Native Russian - Native

Certifications
CS50P: Introduction to Programming with Python — HarvardX / edX (2026):
Credential: https://courses.edx.org/certificates/dc36cfa2d2af48b0a06fd017c4384a9b

Websites and Social Links

GitHub: https://github.com/Shershl
LinkedIn: [відкрити контакти](див. вище в блоці «контактна інформація»)