LLM Spreads: End-to-End Financial Data Extraction
Agentic financial extraction prototype. Built with Vertex AI (SFT), Docling & Modal.
Years of navigating complex data landscapes—across both the DoD and the private sector—have taught me that the most impactful models are built on a foundation of rigorous logic and clean execution. While the bulk of my professional history remains behind corporate firewalls and classified environments, this portfolio serves as a transparent look into my technical methodology.
Leveraging my background in Statistics, I use these projects as a proving ground for modern workflows—applying Data Science, Machine Learning, and Agentic AI to solve the types of high-stakes challenges I’ve encountered throughout my career.
Agentic financial extraction prototype. Built with Vertex AI (SFT), Docling & Modal.
Self-correcting multi-agent RAG for audits. Built with LangGraph, Docling & Modal.
ML-driven pricing & forecasting to end sales bottlenecks and accelerate time-to-close.
99% accurate churn model using SMOTENC & feature interaction engineering.
DoD-grade auditability in retail ML: XGBoost pricing & UMAP high-dimensional discovery.