Open to opportunities

Bohdan Pelekh

@bohdanpelekh

Message

I build and deploy on-device computer vision and NLP models for real-world applications.

Ukraine

Message

I'm building and optimizing real-time ML models for mobile devices at Samsung R&D Institute Ukraine, working across object detection, classification, on-device LLM inference, and audio ML integration.

Previously at IMPACT Initiatives, I designed and deployed NLP pipelines for survey and interview analysis, processing up to 10,000 surveys with hundreds of questions. I built Python, SQL, and Azure data systems and automated data validation and anomaly-detection workflows.

I've trained PyTorch models for OCR and object detection, integrated CV and NLP models, and built Flask REST APIs. My synthetic-image-detection thesis achieved 92.14% mean accuracy on the GenImage dataset using global and local image features.

Experience

Work history, roles, and key accomplishments

Current

ML/CV Engineer

Current

Samsung R&D Institute Ukraine

Jun 2025 - Present (1 year 3 months)

Research, developed and optimized models for real-time on-device inference under mobile constraints. Worked with Object Detection and Classification, On-device LLM inference, and Audio ML integration, deploying optimized models in Android applications using TFLite and ONNX.

PyTorch TensorFlow TFLite Onnx Object Detection Classification LLM Audio ML Android Mobile Optimizations

Data Scientist

IMPACT Initiatives

Dec 2023 - Jun 2025 (1 year 6 months)

Designed and deployed NLP pipelines for survey and interview data analysis, processing up to 10,000 surveys with hundreds of questions. Built data processing systems using Python, SQL, and Azure, and automated data validation and anomaly detection workflows.

Python SQL Azure NLP Data Processing Data Validation Anomaly Detection Automation

Python/AI Engineer

Self Employed

Jun 2023 - Nov 2023 (5 months)

Trained deep learning models using PyTorch for OCR and object detection. Researched and integrated computer vision and NLP models, and built REST APIs using Flask.

PyTorch OCR Object Detection Computer Vision NLP Flask REST APIs Model Integration

Education

Degrees, certifications, and relevant coursework

Kyiv School of Economics

Master's degree, Economic Analysis

2025 -

Master's degree in Economic Analysis, focusing on econometrics, statistics, data analysis, microeconomics, and applied economic modelling. Expected completion in 2027.

Ukrainian Catholic University

Bachelor's degree, Computer Science

2021 - 2025

Activities and societies: Multiple projects in CV & NLP domain: Image Inpainting, OCR System, Image Segmentation, Text Classification, etc.

Bachelor's degree in Computer Science. Thesis on detection of synthetic images using global and local analysis, achieving 92.14% mean accuracy on the GenImage dataset.