Batın Örene

I build and study applied AI systems — LLM and RAG pipelines, agentic workflows, and the evaluation methods that tell us whether they actually work. I am a PhD researcher in Computer Engineering at Bahcesehir University, working on retrieval-augmented generation — standard, agentic, and graph-based — with a focus on evaluation, hallucination detection, and responsible AI. Alongside the PhD I have three first-author papers in progress, targeting EMNLP, ACL, and Information and Software Technology, on agentic-AI verification, responsible release of harmful-speech data, and the gap between "executable" and "correct" in solver-backed reasoning. In parallel, I design applied-AI prototypes at Turkish Airlines, turning LLM, ASR, and automation research into working proof-of-concept tools for high-stakes language assessment — including automated CEFR scoring pipelines benchmarked against human graders to catch model hallucination. My background is unusual on purpose. Alongside a First Class Honours BSc in Computer Science (Goldsmiths, University of London), I hold degrees in educational technology and language teaching and years of experience designing high-stakes assessments. That gives me a rare combination: I can build the model, evaluate it rigorously, and understand the human and pedagogical context it runs in. Interests: agentic RAG, LLM evaluation, hallucination detection, responsible AI, AI for education and assessment.