I’m Evan Seitz, a Computational Postdoctoral Fellow at the Simons Center for Quantitative Biology at Cold Spring Harbor Laboratory. My work focuses on interpreting deep neural networks trained on genomic data, with an emphasis on uncovering the regulatory mechanisms encoded in DNA sequence.
I’m the developer of SQUID and SEAM, two explainable AI methods for understanding what deep neural networks (DNNs) are learning from functional genomics data. SQUID uses surrogate models to interpret cis-regulatory mechanisms and was published in Nature Machine Intelligence. SEAM explores how genetic variation reshapes those mechanisms. I recently presented SEAM at the ICLR Generative and Experimental Perspectives for Biomolecular Design (GEM) Workshop with a preprint currently in preparation for journal submission.
Previously, I completed my PhD at Columbia University in the Frank Lab, where I developed geometric machine learning methods to explore conformational landscapes of biomolecules using cryo-EM data. Across both molecular and genomic systems, a central theme in my work is using machine learning — and rigorous interpretation — to make sense of biological complexity.
Before my scientific career, I worked professionally in 2D and 3D animation and design. I still enjoy visual storytelling and bring this perspective to my research, especially when designing figures and communicating results. Outside of science, I enjoy hiking, biking, tennis, and board games with friends and family.
This site serves as a hub for my work, ideas, and ongoing projects in computational biology and machine learning. For more information, you can download my CV, or connect with me on LinkedIn.
PhD in Biological Sciences, with Distinction (Geometric Machine Learning & Computational Biophysics)
Columbia University, 2017–2022
BS in Physics, with Highest Honor (Biophysics)
Georgia Institute of Technology, 2015–2017
BA in Mass Communication
Georgia College, 2005–2009