Computer Science undergraduate at University of Wisconsin–Madison.
I work on language models, representation editing, activation steering, and efficient adaptation. I am interested in LLM training and post-training, multimodal foundation models, and the systems that make them practical.
- State-Aware Adaptation and Control of Language Models — independent research on parameter/state interventions, LoRA, and KV-cache reuse. Research notes
- LLM from Scratch — CS336-based self-study implementation with dense and sparse-MoE variants, tied embeddings, and tested training components. Portfolio completion and extensions include AI assistance.
- Computer Organization Knowledge Graph — undergraduate research design on educational-text processing, entity recognition, and domain ontologies.
Python · PyTorch · NumPy · Hugging Face Transformers · Git · Linux
Outside research, I enjoy calisthenics and gymnastics rings.