Electrical and Electronics Engineer, graduate of Istanbul Medipol University.
I work on Artificial Intelligence technologies.
Electrical and Electronics Engineering, B.Sc. (100% English)
Honor Student
Siemens
Siemens Mobility
Working student in the Fit4Rail program.
ISTLINK - Beykoz, Istanbul
Development and productization of an AI-powered Emergency Warning System (EWS) with Automatic Safety Alert (ASA) mechanism over DAB+ broadcast.
TUBITAK BILGEM - Gebze, Kocaeli
Istanbul Medipol University
District-level flood forecasting for Turkey. Accepted at SIU 2026 (IEEE) — 0.909 AUC with an 8-seed ensemble.
A first in Turkey: AI-powered Emergency Warning System tested at Çamlıca Tower.
High performance RX/TX patch antenna design operating at 1575 MHz band.
A small, focused Python library for building Claude-powered agents.
@tool decorator, agent loop, prompt caching and
permission system built-in. ~2K LOC, 80% test coverage,
MIT-licensed, on PyPI.
Multi-agent LLM agent grounded in 740K Turkish academic sources with IEEE citations. Live demo on HF Spaces.
NEWAn LLM agent that sits next to you through your whole ML pipeline, from data to production. An 11-command CLI, an MCP server, persistent project memory, and an evidence-bound anti-hallucination contract (0/200 reaching the user in live measurement). MIT, v0.9 on PyPI.
Hakan Sabuniş, Yusuf Ünlü, and Mehmet Kemal Özdemir
2026 34th Signal Processing and Communications Applications Conference (SIU), IEEE, Jul. 2026, pp. 1–4. doi:10.1109/SIU71813.2026.11636290
A Temporal Fusion Transformer-based three-branch hybrid model performing district-level flood forecasting for all 973 districts of Turkey under a strict leak-free 1-day forecast protocol. The 8-seed ensemble achieves AUC 0.909 (F1 0.84, Precision 0.88, Recall 0.81).