Undergraduate Student Researcher
Current- Currently leading team of 4 to find novel visual prompting techniques to incorporate temporality awareness in LLMs
- Normalize and scrape large public datasets using Python and Selenium to improve contextual relevancy by 30%
- Curated model compression techniques of pretrained Hugging Face Transformers, reducing inference time by 40%
- Utilized multi-shot data using MATLAB and PyTorch to enable in-context learning and reduce hallucinations by 25%