Murat Unsal | Data Scientist Portfolio

About

I’m someone who loves Data Science with experience developing data-driven and machine learning solutions in risk and fraud detection. With a PhD background in Environmental Science, I bring strong analytical rigor, scientific thinking, and practical problem-solving skills to build impactful models and tools for business efficiency.

Skills

Python

SQL

Machine Learning

Data Visualization

Statistical Analysis

FastAPI

Streamlit

Power BI

Projects

Fraud Detection ML Model

Developed a Python-based model to detect high-risk transactions with ~85% precision, improving fraud investigation efficiency.

Rule Search Engine

Built a centralized engine for risk rule search and analytics, reducing cross-team query workload.

Automation & Monitoring System

Created a Python automation tool to monitor system performance and rule hit metrics in real time.

Skyulf-MLFlow Library

A personal python ML toolkit/library I built to stop copy-pasting the same preprocessing code in every project

Skyulf End-to-End ML Platform (Personal Project)

Designing a no-code ML platform that automates data exploration, model training, and deployment-ready pipelines.(https://www.skyulf.com/)

Experience

Creator and Engineer – Skyulf (2025 Nov–Present) - https://www.skyulf.com/

Backend: Architected with FastAPI, featuring modular pipeline nodes for data ingestion → feature engineering → modeling → evaluation.
Frontend: Developed an interactive React Flow interface to visually compose and configure ML pipelines.
Orchestration: Integrated Celery + Redis for asynchronous job handling and implemented artifact tracking with joblib to ensure reproducible workflows.
Infrastructure: Designed for flexibility and scale — SQLite for local development, PostgreSQL for production, and container-first deployment.
Focused on data sovereignty and vendor independence, making it suitable for regulated sectors like finance, compliance, and risk analytics.
Currently in early alpha, continuously evolving toward an open-source release with community and grant support.

System Analyst – Western Union (2023–Present)

Developed automation tools reducing manual reporting time by 70%.
Built ML models improving fraud detection accuracy (~85% precision).
Designed a rule analytics engine increasing transparency and collaboration across teams.

PhD Research Assistant – Vilnius University (2020–2025)

Applied machine learning to environmental datasets (>10k samples).
Published peer-reviewed research combining data science and environmental modeling.

Background

PhD in Environmental Sciences

Vilniaus Universitetas / Vilnius University, Lithuania

investigated heavy metal pollution in indoor dust from Vilnius schools, linking it to outdoor soil and airborne particulate matter, and assessed associated health risks for children.

It provides the first comprehensive analysis of indoor dust contamination in Lithuanian schools, establishing correlations between outdoor soil, airborne PM, and indoor dust heavy metal levels.

Methodology: : The study combines field sampling, geospatial analysis, multivariate statistics (PCA, PMF), and health risk modeling using existing soil data and newly collected indoor dust samples.
MSc in Environmental Engineering
Namık Kemal University, Turkey (2017–2019)
BSc in Environmental Engineering
Bartın University, Turkey (2009–2013, Honors)

Contact

📍 Vilnius, Lithuania

📧 murath.unsal@hotmail.com

🔗 LinkedIn

💻 Portfolio