Research Scientist · Data Institute for Societal Challenges

Let's work on it together.

I partner with clinicians, geoscientists, public-health teams, and fellow investigators on problems where machine learning and good software both matter — from a first look at your data to a co-authored paper, a joint proposal, or a platform your stakeholders actually use.

Ways to collaborate

3 tracks
01

Research partnerships

For clinicians, radiology and imaging groups, and geoscience labs

You bring a question and data; I bring the modeling. Typical work is image segmentation and classification, risk and outcome prediction, and explainable results that hold up in a paper and in front of domain experts.

  • Medical imaging — CT, MRI, retinal fundus, mammography
  • SEM and FIB-SEM rock imaging
  • Small or messy datasets, including synthetic data generation
  • Co-authored journal and conference papers
Propose a research project
02

Grant partnerships

For investigators assembling NSF, NIH, HHS, or state proposals

I join as PI, co-PI, or Co-I to lead the data science and delivery aims — methods, data management, and a working dashboard as a tangible deliverable that reviewers and program officers can see.

  • PI on a federally funded HHS / SAMHSA dashboard
  • Data and evaluation aims with clear, measurable outputs
  • Dissemination through interactive public dashboards
  • Experience on NSF- and NIH-funded initiatives
Discuss a proposal
03

Platform builds

For agencies, centers, and programs that need a data product

Secure, role-based web platforms — dashboards, registries, and decision-support tools — built in React with interactive Plotly, D3, and Mapbox visualizations, designed with the people who will use them.

  • Surveillance and planning dashboards
  • Statewide registries with review workflows
  • Geospatial maps and community indicators
  • Role-based access and controlled export
Scope a platform

Where I can help

Backed by published work

If your problem looks like one of these, I've worked on something close to it before.

Healthcare AI

Reliable classification from noisy or web-scale medical images

Healthcare AI

Choosing between radiomics and deep learning for a CAD scheme

Healthcare AI

Predicting prognosis and complications from CT

Clinical tools

Interactive decision support that clinicians can use

Subsurface

Estimating rock properties directly from images

Subsurface

Segmenting scientific images without pixel-level labels

Platforms

Turning state and community data into a secure dashboard

Platforms

Statewide registries and research-development tools

Modeling

Where to place sensors and observations

How to start

A short email is enough. These five things help me reply with something useful rather than more questions.

  1. The problem — what decision or question this work should inform.
  2. Your data — type, rough size, and access status (IRB, data-use agreements, or public).
  3. Timeline — a paper deadline, a funding call, or a launch date.
  4. Funding — existing support, or the call you're aiming for.
  5. Success — a paper, a model, a dashboard, or a proposal.