Résumé
Research engineering, agent evaluation, and applied data.
A public-safe summary of my education, experience, and selected technical work. Phone numbers, detailed address, private research artifacts, and application records are intentionally excluded.
Profile
Systems Engineering M.S. student at the University of Pennsylvania with a Statistics and Economics background. Builds practical AI systems that combine agent research, evaluation, data processing, backend engineering, and reproducible experimentation.
Education
University of Pennsylvania
Coursework includes applied machine learning, optimization, big data analytics, simulation, feedback control, statistics for data science, and time-series machine learning.
University of Connecticut
Research & Experience
AI Agent Systems Research Collaboration with UConn Researchers
- Research reliable tool-using agents, runtime intervention, recoverable execution, evaluation design, and closed-loop control.
- Designed evaluation for safety, task completion, structured block feedback, and post-intervention recovery; co-developed a 60-episode benchmark across five task families.
- Integrated online action interception, evaluated five runtime-defense conditions across 300 traces, and performed a clean-seed leakage audit.
- Co-developed closed-loop trace-replay evaluation across 10,933 shared states and 114 matched telecom tasks.
University of Connecticut, Department of Statistics
- Integrated satellite imagery, OpenStreetMap footprints, and UT-GLOBUS data for city- and building-scale analysis.
- Applied spatial statistics and geospatial visualization; built an HPC/Slurm workflow for 413 major U.S. cities.
GBCS — SkyIT Services
- Collaborated with frontend and backend engineers to deliver and release the Voop application for a client.
- Developed, tested, optimized, and maintained REST APIs with Express.js and Django REST, using Postman for endpoint validation.
- Supported a Firebase-to-MySQL migration through schema refinement and consistency checks.
Beijing QIANSHIDU Trade Co., Ltd.
- Analyzed more than 1 GB of two-year business data and developed a GPT-based demand-forecasting workflow using SHAP and K-means.
- Produced insights that supported reallocating approximately 10% of product supply toward neighboring-country markets.
- Represented the company at Vision Expo 2024 in New York City.
Selected Projects
New Haven Spatial Analysis
Spatial statistics and interactive visualization over 331,423 UT-GLOBUS building records using GeoPandas, PySAL, Plotly, and Quarto.
Evidence-Governed Career Operations Agent
Policy-governed extension of Career-Ops, an open-source project by santifer, using explicit workflow state, evidence routing, approval gates, Playwright, and LaTeX.
NYC Rodent Inspection Analysis
Public-data cleaning, statistical modeling, and geospatial visualization presented to the NYC Open Data team.
NBA Salary Prediction
Feature engineering, cross-validation, Bayesian optimization, and ensembling across four machine-learning models.