Raleigh, North Carolina, United States
I lead an applied AI/ML team that sits at the intersection of research, engineering, and impact. As such, I care deeply about building AI/ML applications that are useful and delight customers while being secure and reliable. This focus on integrating technical innovation with user-centered design is pervasive in everything I do. Whether I'm pioneering LLMOps infrastructure, completing customer beta tests around our latest hybrid search feature, or designing ML or agentic applications to automate repetitive processes, I thrive in making a difference in creative ways. How I work: - I aim to make things better than I found it while scaling value for those around me. - I listen to understand and always aim to co-build with feedback. - I quickly identify barriers and work cross-functionally to make things happen. - I experiment in fast iterations to demo concepts early. - I constantly seek hands-on learning opportunities to find new ways to solve problems. These days, I'm most interested in strategies to scale agentic systems reliably (e.g. custom evaluation frameworks, differential privacy, RAG / GraphRAG, guardrails). Looking forward to connecting!
• Advised graduate students on efficient Hadoop and PySpark coding solutions for big data processing and distributed ML. Encouraged effective problem-solving and debugging strategies to develop early career data professionals
• Derived a reusable interrupted time series pipeline (Facebook Prophet) to identify revenue and user impact of non-attributable NBA marketing efforts, which influenced an engineering culture within the Marketing DS team
• Pioneered a path to production program for deploying machine learning models and monitoring systems (AWS/MLFlow). Reduced retraining times by >80% and captured data/model issues prior to impact on business KPIs while mediating cross-functional communication (Stakeholders/DE/DS) and encouraging MLOps best practices • Modernized account planning of a $1.85B customer segment by building a Monte Carlo CLV model and designing automated Airflow pipelines to run batch backtest/dev/prod model predictions (AWS S3/Redshift). Reduced workload for financial analysts and more accurately allocated product quota across sales reps while generating novel processing techniques to satisfy business requirements • Designed retention and cross-sell propensity metrics using XGBoost and business heuristics to drive performance management and future partner sales across a $4.19B renewal available business. Led team of 2 data scientists in synthesizing requirements and engineering renewal and customer metrics in an external facing partner dashboard
• Modernize existing data infrastructure and ETL processes for an internal financial application using PySpark and SQL • Resolve customer data issues through exploratory data analysis and system domain expertise