Shaz Furniturewala

I work on making language models less broken — fairer, more honest, and better aligned with how humans actually think and communicate.

My research sits at the intersection of NLP and the messier parts of human behavior: how LLMs reproduce bias, how people are persuaded through conversation, how empathy manifests in dialogue, and how content moderation systems fail when pushed. I published at EMNLP, ACL, AAAI, and WASSA in 2024, and I have ongoing work on adversarial robustness in toxicity classifiers and the dynamics of human-LLM conversations on political topics.

I did my undergrad at BITS Pilani, then spent time at the National University of Singapore doing research with the Jaidka Lab. I currently work as an AI researcher at AlphaSense, where I work on applied NLP for financial intelligence.