kaan hacihaliloglu_
AI Engineer and researcher focused on cross-lingual mechanistic interpretability. I study how safety and refusal behavior are represented across Turkish and English LLMs, and build synthetic consumer AI systems at SCL.
~current_work
>
Cross-lingual Mechanistic Interpretability
Investigating how refusal behavior and safety representations differ across Turkish and English model internals
>
SCL Synthetic Consumer Lab
AI Engineer building synthetic consumer systems for realistic behavior simulation and market research workflows
>
MSc Data Science — Sabancı University
MSc thesis on cross-lingual refusal direction analysis in large language models