Nixel

Developed a n-dimensional voxel procedural content generation framework.  Uses a layered approach, where “layers” are modularized operations that can be performed on a substrate content object.

Used to generate maps for AI agent controlled simulated fixed wing aircraft to fly missions in.  Maps included terrain, water features such as rivers and lakes, mountains, different types of biomes such as deserts and tundra, buildings such as airports and ranger stations.

Awarded patent number US-11880930-B2 under “A System and a Method for Content Creation”

Link to presentation: LINK

Klarigo

Tool for explaining AI behaviors, developed for use in the COGLE project for the DARPA eXplainable AI program.

Worked as a full stack engineer, implementing a Django backend to query a PostgreSQL database run in Docker and a React web application frontend.

Designed database model and curated AI traces.

Designed and implemented UI and map renderer.

HYDRA

Hybrid AI/Planning framework for dealing with novel scenarios outside of the AI’s model.

Integrated Neural Network AI output with the domain environments – Angry Birds (Science Birds: https://researchportalplus.anu.edu.au/en/publications/aibirds-the-angry-birds-artificial-intelligence-competition) and Minecraft (Polycraft: https://www.polycraftworld.com/)

Refactored software architecture to make more modular.

Link to paper: https://www.sciencedirect.com/science/article/abs/pii/S0004370224000973

MAVSim

Python wrapper for Ardupilot UAV control system and simulator.  Extends functionality with the ability to save, load, and replay flights stored in a Docker PostgreSQL database.

Later further extended to be a discretized simulator compatible with Nixel generated maps.

Created for use in the COGLE project for the DARPA eXplainable AI program.