PhD student at the Kahlert School of Computing, University of Utah
I’m a PhD student at the Kahlert School of Computing, University of Utah, advised by Ana Marasović in Utah NLP. Right now I’m working on mechanistic interpretability of language models, specifically chain-of-thought faithfulness: whether the steps a model writes down describe computations it actually performed, and whether they show the full path to its answer.
Before Utah, I did my B.S. in Computer Science and Engineering at The Ohio State University. With Yu Su and Huan Sun, I helped build the Mind2Web 2 benchmark (NeurIPS 2025): I wrote tasks and built the web cache that makes its evaluation reproducible. At the Imageomics Institute, I collected and organized the 220M-image training data for BioCLIP 2. I also led perception for the Buckeye AutoDrive Challenge team.
Utah NLP (Ana Marasović), University of Utah
I work on mechanistic interpretability of language models, currently chain-of-thought faithfulness: whether a model's written reasoning steps describe computations it actually performed.
NLP Lab (Yu Su and Huan Sun), OSU
I helped build Mind2Web 2 (NeurIPS 2025 Datasets and Benchmarks), a benchmark for long-horizon web agents: I wrote about 15% of its tasks, ran the Deep Research systems for evaluation, and built the cache of visited pages that makes runs reproducible.
Buckeye AutoDrive Challenge
I led perception for Ohio State's team in the SAE AutoDrive Challenge II, where universities turn a Chevrolet Bolt into a Level-4 autonomous vehicle. I rebuilt the ROS 2 perception pipeline, deployed 3D multi-object tracking, and mentored new perception members in the 50+ person team.