Silas Brack
I'm an AI Research Engineer at Teton, where we're building next-generation healthcare technology to support nurses and patients in hospitals and nursing homes. Our work leverages computer vision and deep learning across several cutting-edge research areas: 3D reconstruction, human mesh recovery, action recognition, and visual-language models. A lot of my work revoves around taking ill-defined business problems, understanding what the essential constraints and requirements are, and using data and statistical models to solve these problems under these constraints.
Previously, I spent over four years at Saxo Bank as an ML Engineer in the Predictive Modelling and AI team under Jeppe Reitz and Nicolas Palm Remy-Perez. I worked on various client-facing projects spanning NLP, representation learning, recommender systems, and traditional machine learning models.
I hold a master's in Mathematical Modelling and Computation from the Technical University of Denmark. My thesis was focused on implicit matrix algorithms for the Laplace approximation in Søren Hauberg's Cognitive Systems group at DTU Compute. My work in the group also led to a publication at NeurIPS based on applying the Laplace approximation to metric learning.
My academic interests are broad within machine learning, statistics: from deep learning and recommender engines to Bayesian inference to network science and forecasting. I particularly enjoy the intersection of theory and practice, and spend much of my time coding and solving mathematical problems. Beyond technical work, I'm passionate about science communication and how research intersects with society. Teaching and effectively communicating complex ideas is something I quite enjoy.
Outside work, I like reading about history and paleontology. I try to make the most of life through continuous learning, spending time in nature, and appreciating beauty wherever I see it. You'll often find me coding on side projects, reading, watching films, listening to music, or experimenting in the kitchen.