Krystal Gong

Hi! I’m Liancheng (Krystal) Gong, a Ph.D. student in Information Studies at the University of Maryland, College Park.
I’m broadly interested in trustworthy large language models (LLMs) and computational social science—building systems that reason and plan reliably in the wild, and using data-intensive methods to understand how people and information move through online ecosystems. My work spans LLM↔formal systems interfaces (e.g., turning model outputs into verifiable planning specs) and large-scale social data analysis (from political communication to media narratives).

Research interests

• Trustworthy & Reliable LLMs: evaluation beyond benchmarks, controllable generation, verifiable planning (e.g., PDDL formalisms), and robustness under feedback loops.
• Computational Social Science: measurement, annotation pipelines, and modeling for political communication, media bias, and platform dynamics.
• Scalable Data Engineering: ETL on large, messy streams; embeddings and representation learning for downstream tasks.

A bit more about my background

At UMD, I’m advised by Julia Mendelsohn and Wei Ai. Previously, I worked on LLM-to-planner pipelines at Drexel (CS), and coordinated large-scale political media analyses at Penn (Political Science)—including scraping and modeling millions of social posts, building annotation systems that align well with human labels, and studying symbolic rhetoric and campaign messaging. I received my MSE in Data Science from the University of Pennsylvania and my BS in Data Science from NYU.

What I’m excited about now

• Making LLM outputs structurally faithful to downstream reasoning modules.
• Building annotation + retrieval + modeling loops that are data-efficient and transparent.
• Turning social-scale datasets into causal, actionable insight—without losing nuance.

If any of this resonates, whether you’re into trustworthy AI, planning, or social data—feel free to reach out. I’d love to chat.