Kyle Harrington 🤖🧪
Kyle Harrington

Staff Research Scientist, AI for Science

About Me

I’m a research scientist at Biohub building foundation models and ML infrastructure for biological imaging—focusing on 3D methods for understanding cellular structure from microscopy data.

Recent work includes: Led the CZII CryoET Object Identification competition on Kaggle (published in Nature Methods, 1200+ participating teams). Built CryoLens, a generative model for learning interpretable 3D representations from cryo-electron tomography data. Created the copick ecosystem for collaborative annotation and data management in cryo-electron tomography.

I serve on napari’s steering council and have built multiple open-source tools: sciview (3D/VR for ImageJ), SNT (neuron tracing and morphology analysis), and album (reproducible scientific workflows).

Background: Led research teams at Oak Ridge National Lab and Max Delbrück Center Berlin. Advised DARPA programs in lifelong learning and adversarial ML. PhD from Brandeis (computer science + quantitative biology), postdoc at Harvard Medical School.

Download CV
Interests
  • 3D and volumetric deep learning
  • Foundation models for microscopy
  • Open-source scientific software
  • Biological image analysis
Education
  • Postdoctoral Fellow, Pathology

    Harvard Medical School

  • PhD Computer Science

    Brandeis University

  • MA Computer Science

    Brandeis University

  • BA

    Hampshire College

Featured Publications
Recent Publications
(2025). A realistic phantom dataset for benchmarking cryo-ET data annotation. Nat Methods.
(2024). Open-source Tools for CryoET Particle Picking Machine Learning Competitions. bioRxiv.
(2024). Surforama: interactive exploration of volumetric data by leveraging 3D surfaces. bioRxiv.
(2024). Polarity-JaM: An image analysis toolbox for cell polarity, junction and morphology quantification. bioRxiv.
(2023). napari-imagej: ImageJ Ecosystem Access from napari. Nature Methods, 20(10), 1443-1444.
Recent & Upcoming Talks
Recent News