Featured work
Selected work — medical AI first, then applied ML and real-time backend systems.
AI-Based Multimodal Alzheimer's Diagnosis System
A university-led research project I joined through the Department of Computer Engineering at TED University and have been contributing to as a volunteer for about a year, working with a team and in collaboration with neurology and nuclear medicine physicians. The work applies deep learning to multimodal brain imaging (PET + MRI) for early Alzheimer's detection; my part covers 3D preprocessing and model training. It was awarded 1st place at Genç Beyinler Yeni Fikirler 2026 (TED University & Turkish Informatics Association) and is also my graduation project.
The project is still unpublished, so its methodology and results are not shared publicly.
Gatherly — real-time communication platform
A server-and-channel communication platform: users create servers, open channels, talk in real time by text or voice, and manage who can do what. Built end to end — from the WebSocket layer to the interface.
- Live messaging
- Typing indicators
- Voice rooms
- Screen sharing
- Presence tracking
- Roles & permissions
- Friends & direct messages
- REST API on Swagger
Product Analysis & Barcode Scanner
Request pipeline — local LLM behind a validated JSON contractIngredient-safety analyser built on a local Mistral model served through Ollama. The prompt enforces a strict JSON output contract and every response is validated against a Pydantic schema before it reaches the user, so malformed model output is rejected rather than displayed. Input by manual text or barcode / QR scan.
View on GitHub →Spades Multiplayer Game
Server-authoritative turn engine — clients send intents, never stateFour-player online Spades (Batak) over WebSockets, with room management, bidding, trump and follow-suit enforcement. Clients never hold the game state: they send intents and the server decides, so a modified client cannot cheat.
Play the live demo ↗ · GitHub →Also built
Experience
R&D in medical AI, plus a backend and DevOps foundation.
TED University — research project
Sep 2025 — Present · AnkaraContributing to a research project on AI-assisted analysis of neurodegenerative disease, joined through the Department of Computer Engineering. I train 3D deep learning models on brain imaging and work on the preprocessing pipeline, together with a team and in collaboration with neurology and nuclear medicine physicians.
Godetya Yapay Zeka ve Mühendislik
Jan 2026 — Feb 2026 · AnkaraWorked on the medical image preprocessing pipeline of the same research project — brain masking and spatial normalisation with MATLAB/SPM, and data preparation for model training — and documented methodology, experiments and results for the team.
Bisoft Bilgi Teknolojileri
Jul 2025 — Aug 2025 · AnkaraBuilt an automated PostgreSQL primary/standby high-availability cluster on Docker with WAL streaming replication and pg_basebackup provisioning, automating the setup with Bash. Migrated the cluster to Kubernetes using StatefulSets, PersistentVolumeClaims, ConfigMaps, Secrets and headless Services, then integrated it with the company's existing failover components and tested failover end to end — cutting a 15–30 minute manual takeover to under 10 seconds.
Skills
Tools and technologies I work with day to day.
Education & certifications
Let's talk
I'm open to Software / ML Engineer roles in Ankara and remote positions. The fastest way to reach me is email — I read everything.