Profile

Software engineer with 8 years of experience taking cloud applications from idea to production for U.S. government-funded research. Owns systems end-to-end: database design, containerized FastAPI backends, CI/CD, authentication, and user-facing frontends, including infrastructure serving generative AI applications.

Experience

Geometric Data Analytics, Durham, NC

Senior Software Engineer (June 2021 - present)

• Lead Engineer, U.S. Forest Service ClassiFuel Project: Designed full system architecture (FastAPI backends, Panel frontend, MongoDB) for an authenticated web app supporting expert image labeling. Built a cloud-deployed database with credentialed access and automated new-user onboarding via Auth0 and SMTP. Integrated semi-supervised and supervised learning into the backend to support an active learning framework in the frontend. Supervised an internal research effort to evaluate and improve modeled surface fuel density against ground truth data collected by USFS collaborators.
• Lead Engineer, LLM-Assisted Ship Design Project: Containerized RAG and diffusion model backends into microservices with one-click CI/CD to Google Cloud Run, abstracting cloud complexity from other developers. Translated naval architect needs into an authenticated frontend supporting design generation, modification, and evaluation via a chat interface that sped up design workflows from days to minutes.
• Principal Investigator, SMoLDER (USDA Phase I SBIR): Built a FastAPI backend and client-side web application (Panel + Pyodide) for real-time probabilistic radio repeater coverage around wildfires. Coordinated technical development with four internal team members and led customer discovery.
• Lead Engineer, DARPA Ocean of Things: Built CI/CD pipelines for automated testing, containerization, and deployment to a custom AWS GovCloud environment. Created and visualized geometric data products from Kafka-streamed sensor data for a continuously shifting network of floats in the ocean, in a multi-developer codebase. Presented progress to DARPA biweekly and coordinated collaboration with other DARPA performers.
• Onboarding: Tailor a technical training regimen for each new hire covering collaborative software engineering: git workflow, repository setup, Python environments, testing and coverage, CI/CD, and documentation.
• Visualization practice: Teach visualization best practices internally, consult on projects to produce more interpretable standalone visualizations for stakeholders, and give company-wide talks on topics from client-side web app deployment to explanatory visualization.

Data Scientist (June 2018 - June 2021)

• Worked on government projects funded by organizations including DARPA, the Air Force Research Lab, and the Laboratory for Analytic Sciences, on topics spanning computational biology, Internet of Things, military logistics, and cryptocurrencies.
• Developed software in a distributed, agile environment on projects with as many as eight developers. Trained classification models in Scikit-Learn. Built static and dynamic visualizations.

Grants

• Principal Investigator, SMoLDER: Situational awareness, ModeLing, and Decision support Engine for wildfire Response, National Institute of Food and Agriculture / USDA, Phase I SBIR, July 2023 - February 2024.

Projects

Hiveplotlib, Creator and Maintainer (2020 - present)

• Open-source Python package for building hive plots, a network visualization technique. Available on PyPI with online documentation.
• Completed major user API improvements with corresponding documentation rewrites, 2024 - 2025. Current focus: speedups and improvements when using hive plots as an exploratory data analysis tool.

Topological Signal Compression (2021 - 2023)

• Led a four-person research effort to assess a persistent homology-based signal compression algorithm. Available on PyPI with online documentation.
• Presented "Topological Simplification of Signals for Inference and Approximate Reconstruction" at the 2023 IEEE Aerospace Conference (slides); Best Paper award for the Remote Sensing Track (out of 25 papers). Also on arXiv.

Polar Parallel Coordinates Plots (2021)

• Developed a novel visualization technique combining concepts from parallel coordinates plots and hive plots. Implemented in hiveplotlib as of v0.16. Paper on arXiv.

Selected Publications

Full list on Google Scholar.

• Koplik, Gary, Nathan Borggren, Sam Voisin, Gabrielle Angeloro, Jay Hineman, Tessa Johnson, and Paul Bendich. "Topological Simplification of Signals for Inference and Approximate Reconstruction." In 2023 IEEE Aerospace Conference, pp. 1-11. IEEE, 2023. (IEEE Xplore, arXiv)
• Ball, Kenneth, Erin Taylor, Nirav Patel, Andrew Bartels, Gary Koplik, James Polly, and Jay Hineman. "Geometric Feature Prompting of Image Segmentation Models." arXiv, 2025.
• Voisin, Sam, Jay Hineman, James B. Polly, Gary Koplik, Ken Ball, Paul Bendich, Joseph D'Addezio, Gregg A. Jacobs, and Tamay Özgökmen. "Topological Feature Tracking for Submesoscale Eddies." Geophysical Research Letters 49, no. 20 (2022). (doi)
• Koplik, Gary, and Ashlee Valente. "The Parallel Coordinates Plot Revisited: Visual Extensions from Hive Plots." arXiv, 2021.

Technical Skills

• Infrastructure & practices: Docker, Google Cloud Run, GitLab CI/CD, GitHub Actions, Infrastructure as Code, microservices, REST APIs, secrets management, Agile, Auth0, MongoDB, Linux
• Languages & tools: Python, SQL, Bash, Git, Claude Code, GitHub Copilot, Conda, Uv, LaTeX
• Key libraries: Bokeh, Cartopy, Dask, Datashader, FastAPI, GeoPandas, Holoviews, Matplotlib, Numba, Numpy, Pandas, Panel, Plotly, Pydantic, Pymongo, Pyodide, Pytest, Ruff, Scikit-Learn, Scipy, Seaborn, Sphinx
• Working knowledge: AWS, D3.js, JavaScript, Jenkins, Kafka, Kubernetes, MLflow, PostgreSQL, Spark, Terraform, TypeScript, Vitest, Vue.js

Education

Duke University · Durham, NC
M.S. Economics and Computer Science · May 2018

Colby College · Waterville, ME
B.A. Economics, Mathematical Science · May 2016
Summa Cum Laude, Phi Beta Kappa

Phillips Exeter Academy · Exeter, NH
June 2012