About me
My name is Qingyang Liu (Ciara), and I am currently pursuing a Master's degree in Computational Data Science (MCDS) at Carnegie Mellon University 🩶♥️. I completed my undergraduate studies with a double major in Data Science and Economics at the University of Michigan - Ann Arbor 💙💛. My passion lies at the intersection of responsible AI, natural language processing (NLP) systems, and the application of machine learning and AI in software engineering. I am particularly interested in advancing these fields with an emphasis on ethical practices and impactful solutions.
I am actively seeking new graduate roles in data science and machine learning, with a particular focus on NLP and large language models (LLMs). If you have any questions or know of any open opportunities, I would love to connect!
Project Highlights
-
LLMENTS: PyTorch for LLM Analysis
A unified Python framework of simple and composable operations for LLM bias and reliability analysis.
Open-source library to be released soon!
-
PEFT on LLMs for Meeting Summarization
A parameter-efficient fine-tuning framework with LongLoRA and shifted attention window to support long-input summarization tasks.
-
Group-Robust Toxicity Detection on CivilComments
A BERT-based classifier for toxicity detection, with a two-stage training regime to mitigate spurious correlation on sensitive demographic groups.
-
Mitigating Algorithmic Bias in Health Risk Assessment
A re-weighting algorithm to mitigate racial bias caused by the proxy label choice in health risk prediction.
