First intern hire for Abstractive Health. Improved the factuality of the NLP transformer models using a method known as constrained beam search. We measured the success of this method using dependency arc entailment (DAE) and the Stanford NLP parser for entity extraction. Subsequently, I improved the accuracy of the product by fine-tuning the LLM models on an Emergency Medicine dataset through Weill Cornell Medicine.

Conducted clinical workflow analyses across Orthopedics, Urology, and Infectious Disease to identify device inefficiencies and design improvement opportunities. Proposed novel device modifications.

Facilitating Python learning for students by developing and implementing self-created, structured lesson plans. Reinforcing classroom material and accommodating various learning styles for other students in their college-level courses.