Yee Man (Kath) Choi

Waterloo, Ontario, Canada

Greetings! My name is Yee Man Choi (you are also welcome to call me Kath). I am an PhD Student at the David R. Cheriton School of Computer Science, University of Waterloo. I am fortunate to be co-supervised by Professors Freda Shi and Krzysztof Czarnecki. I completed my M.S. degrees at Carnegie Mellon University, where I was advised by Professor Lei Li. Prior to that, I received my BASc of Computer Engineering from the University of Toronto, St. George Campus, with a minor in Artificial Intelligence.

My research interests lie broadly in natural language processing, with a particular emphasis on LLM alignment, reliability, and safety. I am interested in understanding and evaluating large language models, including their behaviors, failure modes, limitations and risks. I aim to contribute to research that advances language models not only in terms of performance, but also in trustworthiness and responsible deployment.

news

Jan 26, 2026	HardTestGen is accepted by ICLR 2026!

selected publications

Hire a Linguist!: Learning Endangered Languages in LLMs with In-Context Linguistic Descriptions

Kexun Zhang, Yee Choi, Zhenqiao Song, and 3 more authors

In Findings of the Association for Computational Linguistics: ACL 2024, Aug 2024

Abs DOI

How can large language models (LLMs) process and translate endangered languages? Many languages lack a large corpus to train a decent LLM; therefore existing LLMs rarely perform well in unseen, endangered languages. On the contrary, we observe that 2000 endangered languages, though without a large corpus, have a grammar book or a dictionary. We propose LingoLLM, a training-free approach to enable an LLM to process unseen languages that hardly occur in its pre-training. Our key insight is to demonstrate linguistic knowledge of an unseen language in an LLM’s prompt, including a dictionary, a grammar book, and morphologically analyzed input text. We implement LingoLLM on top of two models, GPT-4 and Mixtral, and evaluate their performance on 5 tasks across 8 endangered or low-resource languages. Our results show that LingoLLM elevates translation capability from GPT-4’s 0 to 10.5 BLEU for 10 language directions. Our findings demonstrate the tremendous value of linguistic knowledge in the age of LLMs for endangered languages. Our data, code, and model generations will be released to the public. Our data, code, and model generations can be found at \urlhttps://github.com/LLiLab/llm4endangeredlang.
HardTests: Synthesizing High-Quality Test Cases for LLM Coding

Zhongmou He, Yee Man Choi, Kexun Zhang, and 6 more authors

2025
CiteGuard: Faithful Citation Attribution for LLMs via Retrieval-Augmented Validation

Yee Man Choi, Xuehang Guo, Yi R. Fung, and 1 more author

2026