Irvine, California, United States
AI Enablement Associate at PricewaterhouseCoopers (PwC), specializing in AI governance, LLM systems, and production machine learning pipelines for large-scale enterprise classification and automation systems.
Building and governing enterprise-scale AI systems including LLM evaluation frameworks, agentic workflows, and production machine learning classification pipelines.
Engineered a full-stack AI platform using Streamlit, FastAPI, and Azure Form Recognizer to automate tax form analysis, reducing processing time from 10 hours to 2 minutes and boosting classification accuracy by 20% through demographic profiling and NLP, with 99% extraction accuracy across 10,000+ financial records.
Developed predictive ML models in Python to improve transportation needs assessments and infrastructure effectiveness by 35%, analyzed LA Metro and CalEnviroScreen datasets to influence decisions affecting 5 million residents, and authored a 30-page report presented to LA representatives, shaping transportation policy decisions with $50 million in funding implications.
Co-authored seven papers in top ophthalmology journals on AI-driven glaucoma research, automated visual field detection with 30 years of data, improved model reliability by 30% through optimization, and enriched healthcare datasets by integrating NHANES and CDC data for enhanced accuracy.
Curate weekly lesson agendas for diverse skill levels and learning styles, create practice materials for applying data science concepts, and develop a comprehensive evaluation method to track member progress and ensure learning objectives are met.
Developed knowledge graphs from NextJS data with Python and NetworkX, led a team in creating an ML pipeline for IoT data from 500,000+ sensors, improved decision-making by 40% through financial transaction analysis, and uncovered key real estate market trends by scraping and analyzing 10,000+ Zillow listings.
Leveraged Python to clean and analyze customer acquisition and student loan data at Savi, used web scraping to develop a property sell-ability index from Zillow listings, and performed quantitative analysis to uncover significant correlations between SARE Ladies survey demographics and real estate identification.