Zeinab Dehghani

AI researcher specializing in AI safety, interpretability, and multimodal systems. Focused on large language models (LLMs) and instruction-based image editing, with expertise in explainability and transparency of generative AI. Experienced in integrating text, image, and speech modalities through projects such as LLM-powered assistants with Whisper and RAG pipelines. Skilled in building high-performance ML pipelines, GPU-based training, and distributed systems, with over a decade of experience in IT infrastructure. Currently working as a Research Assistant with the National Edge AI Hub and the KTP projects.