Projects

(ongoing) Conditioning a PRNN on microstructure parameters using a hypernetwork

An iterative generative art project where an LLM repeatedly extends a p5.js simulation by one feature at a time, producing increasingly complex visuals.

(unfinished) project monitoring key indicators of AI’s influence

Vectorized, JIT-compiled PRNNs in JAX

Adaptive mixing of physics-based and data-driven models

Facilitated BlueDot Impact’s AI Alignment courses

Mechanistic Interpretability Via Learning Differential Equations

DAISR competition project - 2nd place project.

A graph neural network that predicts full-field microscopic strains while keeping the original constitutive law in the loop, accelerating FE² multiscale simu...

A benchmark testing LLMs robustness against bad fine-tuning data (Hackathon project)

of fiber reinforced polymers with variational autoencoders and bayesian optimization

NeuroEvolution of Augmenting Topologies to optimize foundation beams of structures