George Stein Email & Phone Number

› Managing a team of research scientists to deliver on long-term strategic projects for the bank. › Leading research efforts on GenAI.

Berkeley

› Compiled a dataset of 75 million galaxy images and trained a series of convolutional neural networks on multi-node GPU-accelerated systems, leading to two peer reviewed publications. | PyTorch› Implemented self-supervised learning to utilize unlabelled data and improve image classification performance by 170%.› Created and deployed an image retrieval web app, and released community-adopted ML models. | Streamlit › Principal investigator of data science proposal awarded over $150,000 USD of GPU resources.› Designed a new anomaly detection method and applied it to win a blind anomaly detection challenge by characterizing an anomaly occurring in only 0.08% of 1 million events, outperforming 11 other teams.› Built a neural network emulator to replace a costly calculation, achieving 10,000x speed-up. | JAX› Developed a deep generative model to extract insights from noisy and incomplete high-dimensional time series data, reducing the error of a key metric by 45% from the standard non-ML method. | TensorFlow

Berkeley, California, United States

Toronto, Canada

› Constructed a pipeline for generating simulated observations of our universe using high performance computing systems, achieving 100x speed-up. Published ready-to-use simulation products and integrated them into the workflows of a number of large telescope collaborations. › Developed ML segmentation models (U-net) to accelerate cosmological simulations. › Lecturer/Teaching Assistant for 16 undergraduate level courses, including Computational Astrophysics. Co-supervised 9 students.

Toronto, Canada Area

› GPU algorithm development in OpenCL for the purpose of fast radio burst detection in TBs of microsecond cadence radio data› Simulating the late-time observable effects of non-standard physics in the early universe

Vancouver, Canada

› Creating tools to forecast cosmological signals for future experiments

Melbourne, Australia

› Bayesian inference of the cosmic rest frame using peculiar velocity survey data

Toronto, Ontario, Canada

Focused on developing and running high-performance computing simulations

Doctor Of Philosophy - Phd, Astrophysics

Bachelor'S Degree, Honours Physics & Astronomy, With Distinction