Wright-Patterson Air Force Base (WPAFB) · Summers 2022 - 2025
During the 2025 Wright Scholars Research Program, I worked on a database-backed web interface designed to support researchers in the Aerospace Directorate, specifically within the Hypersonics branch. The goal of the project was to centralize research papers and associated datasets into a single system that allowed researchers to reference prior experiments, compare research conditions, and supplement ongoing work.
The system used a Flask-based web application with a SQLite database backend to store papers, datasets, and metadata. To support large-scale ingestion of existing research, the project incorporated prompt engineering techniques to guide a custom-built large language model in extracting structured information from papers and outputting standardized JSON files. These JSON outputs were used to tag data and establish relationships between datasets based on defined parameters.
My primary focus was on project management, requirements gathering, and iterative web interface design. I worked closely with researchers to understand how they searched for prior work and what information was most valuable to surface, then helped translate those needs into concrete features and development tasks. This involved organizing project milestones, refining requirements over multiple iterations, and ensuring the interface remained usable for domain experts in a field unfamiliar to the development team.
Development followed a feedback-driven process, with regular check-ins and interface revisions based on researcher input. Through this project, I gained experience bridging the gap between domain experts and software systems, supporting AI-assisted data processing workflows, and contributing to the development of a maintainable, research-oriented web application.
MPERPG was a multiplayer leadership training simulation built in Unreal Engine 5 using a client-server architecture. The goal of the project was to create collaborative scenarios where multiple players had to communicate and make decisions together under time pressure. The simulation relied on network replication to keep all players in sync and included cooperative challenges such as a procedurally generated maze and a minefield navigation task.
My work focused on implementing core gameplay systems and supporting the multiplayer functionality that kept player actions, objectives, and environment state consistent across clients. I worked with replicated actors and shared state updates, and spent time debugging issues related to desynchronization and timing in a real-time networked setting. I also contributed to character customization and animation systems to make player roles clearer and interactions more intuitive.
I additionally helped integrate data logging features that recorded key player actions and events into structured CSV files for later analysis by researchers and instructors. The project followed an iterative development process driven by frequent feedback from stakeholders, which meant regularly revisiting requirements, making targeted code changes, and refining systems over multiple development cycles. Through this process, I gained experience working in a shared codebase and translating user and customer needs into concrete technical improvements.
C.L.A.I.M. was an AI-driven conversational NPC system developed in Unity to support cultural and linguistic training in a serious game environment. The goal of the project was to create non-player characters capable of engaging in realistic, spoken dialogue while responding appropriately to user input and context through an AI-powered text-to-speech pipeline.
My primary role focused on character integration and project organization. I worked closely with the customer to understand their goals for NPC behavior and interaction, and helped break high-level requirements into concrete, achievable development tasks for the team. This involved translating non-technical expectations into specific features and coordinating implementation across different parts of the system.
On the technical side, I created character animations manually in Blender and integrated them into Unity to support expressive NPC behavior. I also worked on implementing lip-sync functionality by integrating the Oculus Lip Sync library with the text-to-speech pipeline, ensuring that spoken dialogue aligned with facial animation in real time. This required coordinating animation timing, audio playback, and character state within the engine.
The project followed an iterative, feedback-driven development process, with regular reviews and adjustments based on customer input. Through this experience, I became more familiar with managing evolving requirements, organizing work in a collaborative codebase, and bridging the gap between customer expectations and technical implementation.
Teams in Danger was a cooperative multiplayer search and rescue simulation developed for the Air Force Research Laboratory's Gaming Research Integration for Learning Laboratory (GRILL). The project focused on studying teamwork, communication, and decision-making in emergency response scenarios using a realistic, research-driven game environment.
The simulation placed players in a building collapse scenario where they had to coordinate across distinct roles to locate, stabilize, and evacuate victims under time pressure. The system supported multiple player roles with specialized abilities, dynamic hazards such as fire and electrical obstacles, and a beacon communication system that allowed players to mark victims and hazards in real time.
I worked on core gameplay systems and multiplayer functionality in Unreal Engine 4 using Blueprints. This included implementing role-based mechanics, hazard interactions, and victim handling systems. I built a randomized hazard spawner to dynamically generate challenges, and organized ~20 characters, assigning them randomized animations from ~10 variations to represent different victim states. I designed severity levels with descriptive indicators and implemented a system to randomly assign severity to each character.
Additionally, I contributed to features that supported research data collection, including systems that logged player actions and events to structured JSON files for later analysis. Throughout development, the project followed an iterative, customer-driven process, requiring regular feedback sessions and adjustments to meet research needs. Through this experience, I gained early exposure to collaborative development, version control, and building systems that balance technical constraints with real-world research requirements.
The VisionLab, University of Dayton
As a research intern at the VisionLab, I worked on a project involving data collected from an EEG headset, with the goal of identifying and organizing signals from the sensor with the strongest readings. The project required working with raw sensor data and developing tooling to process and analyze that data reliably.
My work focused on building a Python-based data processing pipeline that ingested EEG readings, grouped data by sensor, and identified dominant signal patterns. I used libraries such as NumPy and pandas to clean, manipulate, and analyze structured data, and Matplotlib to visualize trends and validate results during development.
This role gave me hands-on experience working with real-world, noisy datasets in a research setting, where requirements evolved as insights were discovered. I became more comfortable iterating on analysis code, validating assumptions through visualization, and communicating technical findings to researchers with different backgrounds.
Mathnasium of Centerville
As an instructor at Mathnasium, I worked with students across K–12, covering a wide range of topics from foundational number line arithmetic to advanced high school material and ACT preparation. The role required quickly assessing a student’s current understanding and adapting instruction to meet their individual needs.
I regularly broke down complex concepts into smaller, approachable steps and guided students through structured problem-solving strategies. This included reinforcing core mathematical intuition, addressing gaps in prerequisite knowledge, and helping students build confidence with standardized test-style questions under time constraints.
This experience strengthened my ability to communicate technical ideas clearly to a non-technical audience and to adjust explanations in real time based on feedback. These skills have directly carried over into my software and research work, especially when collaborating with teammates, explaining system behavior, or translating high-level goals into actionable steps.