Andrew Yuan
Hello world! I am a PhD Candidate in Computer Science at the College of Information and Computer Sciences at UMass Amherst. I am co-advised by Przemek Grabowicz and David Jensen and am a member of the SIMS Lab and the KDL Lab. Previously, I received my Bachelors of Science from Northeastern University, where I majored in Computer Science with a concentration in AI and a minor in Mathematics.
My research primarily focuses on understanding and evaluating machine learning systems, particularly large language models. I am interested in developing evaluations that test whether LLMs are actually doing the things we claim they are, and in understanding how choices in experimental design and evaluation can shape the conclusions we draw about these models. I am also interested in interpretable machine learning and in applying interpretable methods to problems in domains such as social media, where understanding what a model has learned and how it reaches its decisions can be especially important.
I believe that machine learning systems perform better when we can understand them, especially as humans become increasingly in the loop in ML usage. Our newest AI systems need to be explainable and align with our values and goals. My research explores how we can create these trustworthy and intuitive systems. If you are interested, or just want to talk, feel free to reach out.
news
| Sep 24, 2026 | Our work on understanding compositional generalization in Transformers has been accepted to the NeurIPS 2026 Evaluations & Datasets Track! We study when models can generalize to new compositions of learned behaviors, from controlled composition tasks to fine-tuned language models on code-generation problems. Preprint coming soon. |
|---|---|
| May 25, 2026 | I passed portfolio and am officially a PhD Candidate in Computer Science at UMass Amherst! I also got my Masters from the university at the same time. |
| May 20, 2025 | I spent the Summer at MITRE in their Bedford location as a Cyber Security Graduate Intern, where I worked on a document intake pipeline combining computer vision, custom parsing, and retrieval-augmented LLMs to automate document processing. |
| May 20, 2024 | Presented Dropout Attacks Paper at IEEE S&P 2024. Travel was funded by S&P travel grant. |
| Jan 17, 2024 | Personal website created. |