New York, New York, United States
Machine learning research scientist advancing the foundations of deep learning and approximate inference to accelerate scientific discovery.
Building deep learning models which generalize robustly to new contexts at minimal computational overhead by shaping the implicit bias of optimization. Developing better surrogate models and policies for Bayesian optimization in latent spaces of deep generative models to accelerate scientific discovery.
Developed numerical methods to accelerate large-scale approximate inference.
Designed a variational inference scheme which provably mitigates the approximation bias of large-scale Gaussian processes used for sequential decision-making.
Implemented and benchmarked machine learning algorithms for cell-type deconvolution. Prototyped a clustering method for time-series RNA sequencing data to predict flares in Lupus patients.
Developed a clustering algorithm in Julia to accelerate the error code configuration of on-board computers of new models at BMW significantly reducing configuration time. Designed and implemented an issue tracking process in Jira for the on-board diagnostics team.