Menglin (Miley) Liu

I am an Assistant Professor of Computational Social Science at The Chinese University of Hong Kong, Shenzhen, and an Assistant Professor (by courtesy) at the School of Artificial Intelligence. I also serve as the Coordinator/Advisor for the MPhil-PhD program at the School of AI and the Coordinator/Faculty Advisor of the Reinforcement Learning & Responsible AI (RAIL) Lab. My research bridges political science, NLP, and AI safety, with a focus on American politics, urban governance, and the measurement of political behavior and public opinion.

I develop scalable NLP pipelines to extract meaning from unstructured political text and audio data. My projects include PoliPrompt, an open-source Python package for prompt optimization and LLM-based classification, and large-scale studies on latent ideology in congressional speeches, emotion and effort in open-ended survey responses, and political language in housing policy debates. I am also a cofounder of SurveyFluency, an AI startup for intelligent survey analytics. My research also explores reinforcement learning, agentic AI systems, and AI safety evaluation.

I serve as a National AI Expert for UNICEF China, where I design responsible AI-assisted evidence synthesis pipelines and deploy autonomous research agents for policy evaluation. I am also the Lead Scientist for the Shenzhen AI Industry Association's AI Agent Safety Evaluation Standard, defining adversarial testing protocols and guardrail frameworks for responsible AI deployment.

My work has been published in journals such as Local Government Studies and Cities, and presented in venues across computational social science, including the Society for Political Methodology and APSA. I regularly collaborate with political scientists, policy organizations, and industry partners to translate complex data into actionable insights.