Oguz-Kaan Öztürk

• Design and development of a full-stack web application for a driving school client using Next.js, TypeScript, Prisma, and PostgreSQL, featuring a REST API, responsive design, student management, instructor management, fleet management, financial tracking, and appointment scheduling with conflict detection.

• Deployed in production, with protected API routes based on role-based access control, server-side input validation, audit logging for administrative actions, and EU-hosted PostgreSQL database for GDPR compliance.

• Independent management of the entire project lifecycle: requirements gathering with the client, UI/UX design, development, testing with Vitest, CI/CD, deployment on Railway, and ongoing maintenance.

• Development of a Python-based ML system for detecting and classifying surgical complications, processing over 10,000 pseudonymized patient data from the clinical information system (CIS).

• Building a complete data preprocessing pipeline: ICD-10 code parsing, BMI calculation, median imputation, outlier treatment, and feature importance analysis using Random Forest.

• Training complication-specific neural networks (TensorFlow) with individually adjusted hyperparameters; handling extreme class imbalance (<2% positive cases) using SMOTE oversampling and class-weighted loss functions.

• Delivery of a CLI-based tool for integration into clinical research workflows under a signed non-disclosure agreement.