Currently working with Loris D'Antoni in the UCSD Programming Systems Lab.

Recent Work

Couple research-based projects from Fall 2024:

• Attempted A* Neural Guided Search on ARC-AGI: Neurosymbolic programming approach to apply an object-level DSL to the ARC Challenge.
• Philosophize That: I visualized document-level embeddings on philosophical texts and trained a classification model to label my own writing's philosophy.

Current Work

Presently focused on two projects:

• Extending the R1-style reasoning loop to novel verifiable domains like chess and predicting outputs from code without running it.
• Taking an entropy-based approach to studying continual learning -- focusing on their efficacy and scalability.

More to come!