Heya! I’m Adit.

I am currently at Sandia National Labs developing geospatial machine learning models for predicting wildfire behavior. To-date I've worked in data analytics and interned in frontend software development, venture capital, and marketing. Across all that is a shared love of making thoughtfully designed products with great people, and hopefully improving the world along the way.
I recently graduated with an MS in Computer Science from Georgia Tech, specializing in Machine Learning. Prior to that, I got a BA in Economics from the University of Chicago. Outside of work, I’m a climber, photographer, national park enthusiast, and avid roadtripper. I'm lucky to have lived in 5 different states and visited 44 of them, but these days you're most likely to find me in Albuquerque, New Mexico.

Read more about me...

scroll-down-arrowscroll-down-arrow

Work Experience

Sandia Labs
Hopper
Digitas
Chicago Ventures
Chicago Booth
Polsky Center
Geospatial and ML Intern @ Sandia National Labs
  • Developed algorithm using satellite imagery to map the ignition and spread of historic wildfires, enabling team to model variables that drive fire growth and rate of spread
  • Created deep learning image models with CNNs (PyTorch) to classify landcover from satellite images, owning the whole process from concept to predictive model
  • Supported team ML modeling tasks by building scripts in Python to pull terabytes of historic geospatial remote sensing data from Google Earth Engine
  • Developed geospatial data visualizations to contextualize our work with respect to electric grid resiliency, historic wildfires, and areas with high winds near grid assets

Python

PyTorch

Google Earth Engine

QGIS

Plotly

Figma

Data Projects

Detecting Credit Card Fraud

How do you detect fraudulent credit card transactions when 99.8% of transactions aren't fraud? This page shows how I approach complex predictive modeling on an open dataset, specifically supervised machine learning on tabular data. My goal is for models to be transparent and easy to understand while maintaining strong performance.

Python

Statsmodels

sklearn

Plotly

Seaborn

Web Projects