Full-stack engineer shipping production fintech systems, and an AI researcher teaching retrieval systems to know when they're unsure.
I build software where correctness compounds — payment flows, decision engines, systems that move real money and can't afford to guess. That's what led me to my current research question: how do you make an AI system say "I'm not sure" before it costs someone money?
By day, I'm a Full-Stack Developer on a money-movement product at CentricDXB, shipping features that handle tens of thousands of API requests a month. By thesis, I'm an MSc Artificial Intelligence candidate at Beykoz Üniversitesi in Istanbul, working on confidence calibration for retrieval-augmented generation in fintech contexts.
I'm currently applying to senior frontend and full-stack roles across Europe and Türkiye, and laying groundwork for a fully-funded PhD.
Building and maintaining full-stack features across a fintech product suite — from API design to database performance to production reliability, working within a cross-functional team of 10+ engineers.
A retrieval-augmented generation system that doesn't just answer — it quantifies how sure it is. CC-RAG scores retrieval uncertainty and maps it directly onto transaction risk levels, aiming to cut hallucination in payment and compliance decisions where a wrong answer has a real cost. Production fintech experience feeds directly into how the risk mapping is designed.
A multilingual RAG system answering financial questions across languages, built to bridge the gap between raw financial documents and plain-language answers.
A GitHub bot that reviews pull requests automatically, flagging issues and suggesting fixes before a human reviewer even opens the diff.
A deep-learning NLP application that detects signals of mental health distress from text, trained and evaluated for real-world reliability.