Seattle, Washington, United States
Software engineer graduating June 2026 from UW (CS + Financial Economics). Built a graph optimizer doing Pareto pathfinding across 4 cost criteria, an NLP pipeline over 100k+ NCI abstracts, and automation tools now used by 15+ engineers at Fabrinet. --> Graduating June 2026. F-1/STEM OPT eligible. Reach me at [email protected]
Led training and onboarding for 20+ technical consultants, redesigning technical curricula and mock ticket systems to improve troubleshooting readiness. Provided Tier-1/2 support for enterprise platforms (Canvas, Zoom, Panopto) used by thousands of UW students and faculty. Investigated software bugs, documented solutions, and collaborated with vendors to resolve service reliability issues.
Managed and revised a training curriculum for new consultants, incorporating new content, consolidating modules, and removing outdated material to enhance efficiency and relevance. Developed and integrated new resources into training, including mock tickets for recent issues and flowcharts for visualizing complex processes, improving trainee preparedness and problem-solving skills. Reorganized and consolidated a task-tracking database (Excel sheet), enhancing data integrity and providing clearer metrics for progress tracking.
Supported UW instructors and faculty by providing responsive technical assistance via email and phone. Led investigations into software issues and worked with vendors to implement effective solutions, improving the performance and reliability of LMS tools including Canvas, Zoom, and Panopto. Researched, tested, and documented technical solutions, gaining hands-on experience in software testing and QA.
Solution adopted by 15+ engineers company-wide to analyze 100+ PCB test results weekly, improving debugging speed and reliability across hardware validation workflows. Developed Python tools to automate PCB testing workflows, including a log file parser for an In-Circuit Testing machine that extracts serial metadata and component measurements, and converts raw results into structured CSVs for efficient analysis. Designed and implemented a Python GUI application with watchdog-based monitoring, retry logic, and color-coded logging to automatically manage test output files, improving automation and usability for engineers handling high volumes of manufacturing data.
Developed a BERTopic model to categorize to categorize 100,000+ National Cancer Institute research grants using their abstracts, improving grant categorization accuracy and aiding in research prioritization. Determined optimal number of clusters (15-240) via elbow method, and graph cost valuations for various grant types to filter out non-significant data. Addressed challenges such as document outliers and duplicate topic naming. Collected and visualized findings using Altair to analyze the allocation and spending rates of funding across different clusters.
Built a Sales Forecaster web app using ML models for a supermarket chain's demand forecasting pipeline. Applied feature engineering and dimensionality reduction to improve model prediction accuracy by 17%.