Machine learning researcher and engineer
I'm pursuing an MSc in Artificial Intelligence at the University of Amsterdam after completing my B.S. in Computer Science at UC Irvine.
I'm particularly interested in probabilistic machine learning and generalization.
Research
Stochastic optimization for reinforcement learning
At UC Irvine's Intelligent Dynamics Lab, I built PyTorch tools for gradient estimation and studied continuous relaxations for discrete world models.
Generalization and model merging
With UC Irvine's Machine Learning Research Group, I investigated double descent, minima flatness, and parameter-efficient ways to merge pruned neural networks.
Machine learning for physical systems
At Calit2, I developed neural-network surrogates for finite-element thermal simulations and real-time computer vision pipelines.
Recent work
Recent projects include a graph variational autoencoder for molecular generation and semi-supervised Python–C++ translation with CodeT5. I have also worked on production systems at Uber and Microsoft.