Kristin Gallik, Ph.D.
Machine Learning & Computer Vision Scientist | Bioimage Analysis Specialist
Who I Am
I'm a computational scientist who transforms complex microscopy data into quantitative biological insights using machine learning and computer vision.
What makes me different is my unique path: I started as an experimental biologist and became a machine learning practitioner—giving me deep expertise in both the structure of imaging data and the computational methods to analyze it.
I don't just apply existing tools; I develop custom solutions tailored to specific biological questions and imaging challenges. Whether it's training deep learning segmentation models, building end-to-end analysis pipelines, or consulting on experimental design, I bridge the gap between what biologists need and what computational methods can deliver.
My Journey: From Microscopy to Machine Learning
​The Foundation: Experimental Biology
I earned my PhD in Biological Sciences (Molecular, Cellular, Developmental, and Neurobiology) from the University of Illinois at Chicago in 2022. My doctoral research involved extensive hands-on microscopy work—live 3D time-series fluorescence imaging, serial transmission electron microscopy, and 3D cellular reconstructions using zebrafish and C. elegans as model systems.
These experiences gave me an intimate understanding of how imaging data is generated, what experimental controls are critical for quantitative analysis, and how data quality determines what can be measured. I learned to see beyond the pretty pictures to understand the underlying biology and technical limitations.
​
The Transformation: Learning Computational Skills
When I joined the Van Andel Institute's Optical Imaging Core in 2022, I was the sole person providing bioimage analysis support for the entire institute. I quickly realized that solving the diverse, complex problems researchers brought to me would require advanced computational skills I didn't yet have.
So I learned—intensively. Over the past 4 years, I've gained expertise in:
-
Python programming (now my primary language) with libraries like NumPy, scikit-image, and pandas
-
Deep learning frameworks for computer vision and image segmentation
-
Transfer learning approaches to fine-tune models for specialized biological datasets
-
Software engineering practices including version control (GitHub), project management (JIRA), and reproducible workflow design
In 2023, I was accepted into DL@MBL (now AI@MBL), a highly competitive, 2-week intensive course on deep learning for bioimage analysis at the Marine Biological Laboratory (University of Chicago). There, I developed a custom CNN U-Net for semantic segmentation of transmission electron microscopy images—my first deep learning model built from the ground up.
​
Where I Am Today
Today, I'm one of two bioimage analysis specialists supporting approximately 300 scientists across 40+ research labs at Van Andel Institute. I've:
-
Developed custom machine learning models for instance segmentation, semantic segmentation, and image classification
-
Built analysis pipelines for digital pathology, 3D confocal microscopy, and transmission electron microscopy
-
Created infrastructure for collaborative science, including a GitHub organization, JIRA project management system, and OneDrive collaborative notebooks
-
Launched monthly office hours where researchers can get experimental design consultation and analysis guidance
-
Trained deep learning models using transfer learning with CellPose, StarDist, and other published microscopy deep learning architectures
What Makes Me Different
I understand imaging from both sides. Many machine learning practitioners don't deeply understand microscopy. Most biologists don't build custom neural networks. I do both.
Key advantages:
-
Biological intuition: I know what's biologically meaningful vs. artifact, what experimental controls matter, and how to design analyses that answer real biological questions
-
Technical depth: I can modify neural network architectures, implement custom loss functions, and build production-quality analysis pipelines
-
Problem-solving mindset: I proactively identify gaps and build solutions—from project management systems to novel ML approaches
-
Educator at heart: I believe in empowering researchers with the right tools and knowledge, not just delivering black-box solutions
Current Work & Expertise
Deep Learning & Computer Vision
-
Instance and semantic segmentation using transfer learning
-
Image classification with knowledge extraction
-
Custom model training and fine-tuning (CellPose, StarDist, U-Net architectures)
-
Multi-modal bioimage analysis (Xenium and Visium HD combined with fluorescence slide scan images)
Imaging Modalities
-
Digital pathology whole slide images
-
3D confocal and multiphoton microscopy
-
Transmission and scanning electron microscopy
-
Light sheet microscopy (SPIM)
-
Time-series and volumetric datasets
Pipeline Development
-
End-to-end Python-based analysis workflows
-
GPU-accelerated image processing (pyclesperanto)
-
Batch processing for high-throughput analysis
-
Reproducible, version-controlled code (GitHub)
Consultation & Design
-
Experimental design for quantitative imaging
-
Imaging modality selection and optimization
-
Collaborate with biostatisticians on power calculations and statistical analysis
-
Training and education for researchers
Featured Project
First Computational Model of Mammary Gland Involution
My current flagship project involves creating the first-ever computational model of mammary gland involution using machine learning on over 1,200 H&E whole slide images. This massive collaborative effort with a graduate student and biostatistician includes:
-
Machine learning pixel classifiers for tissue component quantification
-
Tile-based spatial modeling of tissue architecture
-
Deep learning image classifier with knowledge extraction for novel biological insights
-
Custom instance segmentation for adipocyte morphometry
This project exemplifies my approach: combining multiple ML techniques to solve a complex biological problem that has never been tackled computationally.
What I'm Looking For
I'm actively seeking opportunities to deepen my expertise in machine learning and computer vision for bioimage and multimodal biological datasets. I'm particularly interested in roles that focus on:
-
Developing novel AI approaches for biological imaging
-
Building tools and pipelines that enable biological discovery
-
Working in interdisciplinary teams at research institutes or biohubs
-
Eventually moving into leadership positions where I can guide teams in computational biology
Dream organizations include the Allen Institute, CZI Biohubs, Calico, and similar institutions pushing the boundaries of computational approaches to biology.
Beyond Work
When I'm not training neural networks or debugging code, you can find me learning new recipes, weight lifting (I'd like to qualify for a lifting competition one day!), and hanging out with my dogs and cat.
Let's Connect
I'm always excited to discuss challenging problems in bioimage analysis, machine learning applications in biology, or opportunities to collaborate.
​
GitHub: https://github.com/vaioic
LinkedIn: https://www.linkedin.com/in/kristingallik/
Email: kristin[dot]gallik[at]gmail[dot]com​
Projects: Learn More Here
Technical Skills Summary
Programming: Python (primary), ImageJ Macro, Groovy
ML/DL Frameworks: CellPose, StarDist, BrainGlobe, PyTorch/TensorFlow
Image Analysis: NumPy, scikit-image, pyclesperanto, ImageJ, QuPath, Napari
Tools & Platforms: GitHub, JIRA, Jupyter, VS Code, GPU computing, HPC
Microscopy Expertise: Confocal, Digital Pathology, TEM, SPIM/Light-Sheet, widefield, super-resolution
Formal Training:
-
PhD in Biological Sciences, University of Illinois at Chicago (2022)
-
DL@MBL (AI@MBL), Marine Biological Laboratory (2023)