First full-time engineer for Duckie, an AI agent for customer support teams. Built generative AI solutions using OpenAI, Anthropic, and Bedrock APIs, improved RAG accuracy by 248%, slashed token usage by 50%, and shipped 20 full-stack features.

Contributing to SWE-agent, an open-source Software Engineer agent, by building improved semantic search functionality for faster autonomous coding. Collaborating directly with OpenAI on SWE Bench Verified.

Summer Enhancement in Earth Sciences (SEES) Program. Researched impact of solar arrays on agrivoltaic farming performance.

Co-authored 3 papers on feasibility of using space solar power systems (SSPS) to supply and balance electrical grids across different U.S. climate regions. Presented at NASA, IEEE, and the European Space Agency.

Created a rocket thrust optimization model and published 4 first-author papers. Accepted to a global flagship aerospace conference and attended in Rome.

Open-source software with 1M+ views. Recognized by PhDs, NASA scientists, Oxford biophysicists, and "Boids" algorithm creator Craig Reynolds.

Software tool that determines the most optimal prompting method, from zero-shot to three-shot, for a given task using advanced ML techniques.

Tool generator for LLMs and AI agents to create customized toolsets, enhancing their capabilities and task performance.

Lightweight project that converts unstructured data to structured markdown format, optimizing content for LLM prompting.