Data scientist with a physic background working at the intersection of metrics R&D, model development, model implementation, and data engineering. I am passionate about finding stories hidden in data and ensuring statistical soundness during model design.

As a PhD student and Postdoc I developed statistical models, computer-vision algorithms, physics simulations, and data-analysis pipelines to answer fundamental questions in biophysics and medicine.

Outside of work I enjoy playing trombone with the NYC Googler Orchestra and the Bayonne Bridgemen Brass, playing boardgames, and doing fun activities with my wife and dog.


Data Scientist (current)
Google Maps
Oct 2021 -
New York, NY

Data Scientist
Credit Modeling and Analytics
Jan 2021 - Oct 2021
New York, NY

Postdoctoral Researcher
Institute for Physical Science and Technology
Aug 2020 - Dec 2020
College Park, MD (remote)


PhD in Physics, Aug 2020
University of Maryland College Park
College Park, MD
Dissertation: Quantifying the Organization and Dynamics of Excitable Signaling Networks

B.S. in Physics, May 2013
St. John's University
Jamaica, NY
Minors in Mathematics and Chemistry


Monroe H. Martin Fellow
(Continue reading)

(Continue reading)

Department of Physics/CMNS
Outstanding Graduate Assistant
(Continue reading)


Modeling Flow-Field Dynamics in Biological Systems

Optical flow measurements enable a wide variety of technology, from facial recognition to self-driving cars. In this project, I designed an optical-flow-based pipeline to quantify flow-field dynamics in biological systems.

By applying my algorithm to microscope images, I obtained measurements for things like speed and directionality in a variety of cell types. I then built a model of these flow fields with a bimodal mixed von-Mises distribution to parameterize the effect of external forces on the cells.

This algorithm was introduced in Lee*, Campanello*, at al. MBoC 2020 (preprint and manuscript) and will be featured in several upcoming publications.

The original MATLAB code for the MBoC manuscript can be found on github at, and the updated version at

Key concepts: computer vision, optimization, modeling, maximum-likelihood estimation, data visualization, time-series analysis.

Applying Excitable-Systems Models to Biological Signaling Networks

Excitability is the phenomena whereby a system can rapidly change states, e.g., the switching between a flat and oscillatitory state in the red line on the movie to the right.

In my doctoral dissertation, I explored the role of excitability in biological systems, and identified key mechanisms that enable excitability during physiological processes such as immune response and wound healing.

One excitable-systems model that demonstrates this concept is the FitzHugh-Nagumo model, an implementation of which can be found on my github:

Key concepts: nonlinear dynamics, nonlinear systems, coupled system, mathematical modeling, time-series analysis.

Quantification of 3-D Filament Networks.

Filament networks are ubiquitous in biology, from the extracellular matrix that cells use to navigate their environments, to the wiring of the nervous system with axons and dendrites.

In this project, I developed a robust and automatic algorithm to segment (i.e., extract), skeletonize, and disentagle filament networks based on their local topological features.

Some of this work is featured in Campanello, Traver, at al. BioRxiv (2020) and will also be utilized in upcoming papers in preparation.

Simulations of filament-network polymerization and depolymerization

The reaction-diffusion dynamics that mediate biological polymerization and depolymerization are of great interest for their role in cell regulation.

In this work, I utilize an excitable-systems and reaction-diffusion-like model to simulate the polymerization and depolymerization dynamics of Bcl10, a key intracellular protein in T cells.

In this work, I utilize an excitable-systems and reaction-diffusion-like model to simulate the polymerization and depolymerization dynamics of Bcl10, a key intracellular protein in T cells.

Featured in Campanello, Traver, at al. BioRxiv 2020.

Modeling and Visualizing Crime in New Carrollton, MD

I coached a team in the 2019 UMD Data Challenge where we built a model and visualization tools to help police officers in New Carrollton, MD predict where crime and other police indicents are likely to take place.

Using 5 years of recent data, our model identified locations and times of day in which crimes were most likely to take place. We overlaid the model's outputs on a map of New Carrollton, shown here.

This project was awarded by the AWS judges for "Best Presentation."


MATLAB Boot Camp (2015-2019)

From 2015 to 2019, I organized and taught an annual week-long MATLAB Boot Camp on image processing, computer vision, statistical modeling, and data analysis in MATLAB. More than 150 students and researchers have attended the boot camp since 2015, including graduate students, postdocs, and PIs, many from the nearby National Institutes of Health and Johns Hopkins University.

Class of 2017

UMD Data Challenge (2019, 2020)

In 2019 and 2020, I coached teams of data-science students to address data-driven problems for a university-wide competition. In 2018, we built a model that visualized and predicted crime in New Carrollton, MD based on data from the local police department; and in 2020, we did statistical analysis of traffic data to help inform Maryland and DC Departments of Transportation on how they can alleviate traffic congestion.

DC 2020 team

COMBINE Committee (Co-chair 2018, 2019)

In 2018 and 2019, I co-chaired the COMBINE Committee in charge of program management for the COMBINE NRT ( to develop and organize their extracirricular programs, including outreach, research programs, internship fairs, and career-development workshops.

We organized two Data Science Career workshops attended by more than 300 graduate students and postdocs at the University of Maryland.

Furthermore, with network science being a core theme in COMBINE, we also organized several network-science- and data-science-themed Maryland Day demonstrations.

GRADMAP Winter Workshop (2018, 2019)

GRADMAP is a graduate-student-led program to promote graduate studies in physics and astronomy to underrespresented groups. In 2018 and 2019, I taught and mentored students as part of the "Winter Workshop", which is a 10-day-long workshop where late-year undergraduate students learn python, develop a simple research problem, and deliver a final presentation on how they addressed their problem.

One of my students was interested in physics and built a simple simulation of quantum tunneling in periodic square wells. Another was interested in biophysics, and tried to quantify a fundamental problem in cancer biology: individual vs. collective behavior in metastatic cancer.

Teaching quantum mechanics

My mentee, Akorede, delivering his final presentation

"Can you move like a cell?" (Maryland Day 2016-2019)

Collective motion is an important biological process in cells of all types. For example, skin cells move collectively during wound healing, and immune cells work collectively to detect and eliminate infections.

For the annual Maryland Day, I designed an interactive demonstration that invited participants to "move like a cell" as a series of webcams measured their motion in real time using algorithms such as Crocker-Grier particle tracking, and Lucas-Kanade optical flow. Participants were scored on how collectively they move while running around and playing games like follow the leader.

The software I wrote for the demonstrations can be found on my github.
Particle tracking of red hats: (code)
Live optical-flow-based analysis of motion: (code)

Results of the code

Explaining the data

Directing traffic

Losert Lab