I'm Mehmet.

• Deliver end-to-end features for a developer IDE platform, spanning Node.js services, RESTful APIs, and React/TypeScript UIs over a real-time WebSocket sync layer.
• Led the front-end team on the developer IDE platform project.
• Contribute to HarmonyOS app development across smartphone and smartwatch (wearable) devices, adapting UIs to constrained form factors.
• Improve data synchronization with better serialization protocols and optimize the performance of large, virtualized list rendering in data-heavy views.
• Design reusable, component-based architecture and set up monitoring and analytics for key user flows, working in an Agile team with sprint planning and code reviews.
• Previously contributed to Petal Search web and mobile experiences with Vue.js, resolving critical issues and improving search-engine performance.

• Built a multi-site CMS and university management systems from scratch, end to end (.NET backend, SQL data models, Vue.js UIs), managing the full project lifecycle.
• Developed and maintained ~100 web applications and internal systems supporting academic and administrative workflows.
• Designed and integrated RESTful APIs; gathered stakeholder requirements and translated them into modular frontend features.

• Built an accounting web application with Django and Vue.js, still in active use with low maintenance cost.
• Developed native and hybrid mobile apps with Swift and Ionic (Angular); created a real-time meeting app with Java and Socket.IO.

• Contributed to an ESB (Enterprise Service Bus) middleware project in Java enabling service-to-client interaction.
• Developed key modules for a BI tool using WPF and .NET for data visualization and reporting.

• Developed ERP modules using Java, .NET, jQuery, and MSSQL; built hybrid mobile apps with Ionic/Angular.
• Maintained and enhanced high-traffic websites, including www.turkishairlines.com.

A multi-tenant, multilingual (TR/EN) content management system that runs the university's main website, faculty and unit websites, and academic staff profile pages from a single backend. A new unit site goes live from the admin panel by choosing a domain and a template, with no code change.

• Backend: ASP.NET Core Web API in a layered 5-project solution, EF Core + MSSQL, Dapper, Autofac, AutoMapper, JWT; 73 REST endpoints across 17 controllers and 18 entities.
• Frontend: 5 Vue 3 + Vite SPAs (visitor site with 4 templates, site admin, global admin, personal pages) using Pinia, Vue Router, vue-i18n and a custom CKEditor 5 build.
• Highlights: domain-based site resolution, recursive menu tree and mega menu, drag-and-drop ordering, enterprise SSO integration, and a nightly background service that synchronizes staff data.

Full-stack web projects with Django and PostgreSQL, containerized with Docker and deployed on AWS using GitHub Actions CI/CD.

Thesis on predicting thyroid nodule malignancy risk with graph neural networks. Completed alongside full-time engineering roles.

Applied deep learning to healthcare data, comparing recent deep learning models on detecting malaria parasites in red blood cells, and built a mobile app exposing the models through intuitive UIs.

M. Yavuz and N. Yumuşak, "ROI+Context Graph Neural Networks for Thyroid Nodule Classification: Baselines, Cross-Validation Protocol, and Reproducibility." doi:10.3390/electronics15010151

• Languages: TypeScript, JavaScript (ES6+), Python, C#, Java, SQL
• Frontend: React, Vue.js, Angular, HTML5, CSS3/SCSS, styled-components, responsive SPA development, virtualized rendering
• Backend: Node.js, .NET, Django, RESTful APIs, WebSocket real-time sync, microservices & ESB middleware
• Databases & Cloud: PostgreSQL, MSSQL; Docker, AWS, CI/CD (GitHub Actions), Git
• Machine Learning & Research: graph neural networks, deep learning (ResNet-50 transfer learning), medical image analysis, cross-validation protocol design, model calibration
• Practices: end-to-end feature delivery (schema to pixels), Agile (Scrum), code reviews, testing, monitoring & analytics, performance optimization