Hi, I'm Andrew Sumner
Software Engineer, currently working in fintech. I build tools for the communities I'm a part of.

Selected Highlight Project and Corresponding Design Documents
Parkscout.io - National Park Mapping Tool and Trip Planner
This ongoing project is intended to be a one-stop shop for all the data and tools needed to plan a trip to a U.S. National Park.
About Me
Software Engineer with 4+ years of experience across fintech, quantitative finance, and banking. I specialize in full-stack development with strong backend expertise, utilizing Python, TypeScript, Kotlin, Ruby/Rails, and AWS.
Currently at Upstart, I focus on borrower verification systems, enhancing data accuracy and approval rates through features like new credit reporting integrations and customer-facing UI improvements to minimize friction. Previously at Capital One, I designed and built high-impact products including a real-time marketing decision platform handling 100+ million daily records and a self-service cashier's check kiosk deployed across multiple locations. I also spent a short time at Campbell & Company, where I automated analytics and data workflows used by quants.
With a background in both computer science and physics, I bring strong analytical and problem-solving skills to each project, helping me navigate complex challenges across the full development lifecycle. I'm passionate about bringing practical, robust solutions to life.
In my personal time, I love staying up-to-date with new technologies and am currently building a web app focused on outdoor recreation, powered by large language models. If you're interested in connecting, let's chat!
Experience
Software Engineer, Borrower Verifications at Upstart
May 2025 - Present
Quantitative Software Engineer at Campbell & Company
December 2024 - April 2025
Senior Associate Software Engineer at Capital One
August 2021 - October 2024
Education
University of Maryland, College Park
BS in Computer Science & BS in Physics, 2021
Minor in Technology Entrepreneurship • GPA: 3.52
Skills
Programming Languages
Frameworks & Libraries
Tools & Platforms
Other Skills
Certifications
Featured Projects
parkscout.io - National Parks Explorer
Interactive web application for visualizing and exploring U.S. National Parks data. Ongoing project.
Park Visitation Analysis
Data analysis and visualization of National Park visitor patterns over time, including seasonal trends and impact of various factors on park popularity.

Land Use Prediction
Built a convolutional neural network using TensorFlow to classify land use from satellite imagery with 85% accuracy on test data.

Panorama Image Stitching
Developed complete image stitching solution implementing advanced computer vision algorithms including SIFT features and RANSAC.

Galaxy Classification
Created CNN using PyTorch to classify galaxy types from 18,000 images, achieving 70% accuracy despite hardware limitations.