- Research regarding interpretability and explainability of deep learning models- Investigated the introspective power of LLMs (Paper Written + Submitted)- Exploring Transformer and S6 interpretability through causal tracing- Project converting driving manuals into interpretable triples of tokens to feed into autonomous vehicles

- Working on classification model behind OpenTutor (learning framework)- Obtained and curated massive dataset of all OpenTutor responses- Finetuned GPT and LLaMA (with QLoRA) to perform better on custom data- Implemented custom metrics to measure performance

Lecture weekly to dozens of students for 60-90 minutes on the field of AI. Create dozens of quality lectures, exercises, challenges, and demos to guide and instruct attendees on the fundamental theory, mathematics, and practical use of deep learning. Contacted 25+ companies requesting sponsorships and funding.