• Conducting research on cross-functional teams for LLM Red/Blue teaming and Binary Function Similarity Detection

• Spearheaded 2 research projects in Computer Vision through a full lifecycle (ideation, development, maintenance)TernaryMixOEA new training function and two evaluation methods for Computer Vision Out-of-Distribution (OOD) detection yielding 18-point improvement in AUROC GCNN/OOD Viability Study Piloted new research into the use of cutting-edge AI architecture in OOD detection, with early results indicating 20-point growth in AUROC and a 41% improvement in False positive detection compared to existing architectures• Publication: “Establishing New Baselines and Introducing TernaryMixOE…” (Fleischmann, et al.) Pending review • Presentation: Invited talk at the AFRL Center of Excellence to present research findings

• Created, installed containerized CI/CD data integration pipelines to connect large-scale customer data to Cognite’s data contextualization platform• Wrote, documented, maintained production Data Analysis software for 6 clients in cross-functional teams leveraging Amazon Web Services (AWS) and Microsoft Azure. Projects include:Data Ingestion and AI Generation for Power and UtilitiesDesigned, developed pipelines to ingest power grid reliability data to train a Generative AI model for client demosVirtual Flow Meter ML DashboardDesigned, user-tested a real-time dashboard with Plotly Dash to visualize 7 ML virtual flow meter algorithmsPharmaceutical Manufacturing Error-detection Feasibility StudyUsed Natural Language Processing to identify errors and defects in manufacturing processes in an introductory feasibility study for 35 lines across two plants for a major pharmaceutical manufacturerEmissions Reporting ModernizationWorked in an AGILE team as a data engineer to reorganize the emissions calculations and reporting process for an oil and gas company, resulting in an optimization of 2000 man-hours per year.

• Conducted original research in Computational Knot Theory, including establishing new visual and numerical representations of Torus Links• Presented results to faculty in the department and wrote our findings in a final report• Condensed existing research into digestible write-ups designed for the early undergraduate level