Pioneering mechanistic understanding of biological AI models

AI in digital health using wearable data to connect sleep regularity and phenotypic data to mental health outcomes such as depression.

AI for protein design in Tommi Jaakkola and Regina Barzilay’s Lab building non-autoregressive graph based sequence models for the protein sequence recovery task i.e. inverse protein folding.First author of "Fast Non-autoregressive Inverse Folding with Discrete Diffusion" at NeurIPS Machine Learning for Structural Biology Workshop 2024.

Quantitative Strategist Intern.First rotation on Equities Structured Products desk calibrating stochastic volatility model for exotic path dependent options. Second rotation on Electronic Market Making researching ETF pricing signals.

Working on the Computational Statistics department. Researched utilizing feature attribution algorithms on deep neural networks to identify significant genes for classification between diseased and healthy cells using single-cell RNA-seq data.

Worked with supervised behavioral ML models using BORIS and DeepLabCut studying depression resilience in submissive mice. Authorship on "Behavioral and dopaminergic signatures of resilience" published to Nature.Shadowed for eight months for 250 hours. Familiar with live electrophysiological recordings and optogenetic stimulation for circuit analysis between the mPOA and VMHvl. Investigating the rewarding neurological properties of testosterone by performing a testosterone conditioned place preference (CPP) correlated with aggression CPP in mus musculus.