Track 01 — Data Science & AI/ML
Production AI — not proof-of-concepts.
From LLM fine-tuning and RAG pipelines to enterprise NLP automations processing nine-figure transaction sets — shipped in regulated environments where failure costs real money.
State & Local Tax AI Classification Engine
The Opportunity
State and local tax jurisdictional classification was manual, inconsistent, and generating costly purchase-order errors at enterprise scale — with hundreds of tax-scenario permutations no analyst could reliably track.
The Solution
Co-led development of an AI-driven SALT classification engine using Python ML/NLP and LLM reasoning. Built a chatbot interface modeling all jurisdictional scenarios with interpretable outputs and an automated classification pipeline.
The Result
415,000 tax scenarios modeled. Targets a 31% reduction in PO error rates through automated jurisdictional classification — turning a recurring high-cost manual process into an auditable AI workflow.
Recognition
"
Selected at the Lockheed Martin internal AI Hackathon for demonstrating enterprise-scale AI capability applied to a real, high-cost compliance problem. Leadership recognized the business impact potential and the depth of the AI engineering approach.
AI Hackathon · 2024 · Leadership Selection
Hackathon — Action Shots
Production Withholds Automation — Enterprise Financial Intelligence
The Opportunity
A nine-figure withhold balance on a critical defense program had no automated visibility. Analysts spent thousands of hours manually pulling, reconciling, and reporting data — with no scalable path to leadership-actionable insight.
The Solution
Built a full Alteryx ML/NLP pipeline (356-tool workflow) extracting from SAP HANA and program systems, with Python anomaly detection and SQL integration feeding a Tableau executive dashboard on SharePoint — fully automated, refreshed on schedule.
The Result
2,132 annual labor hours saved. Low-six-figure losses prevented. Adopted into standard business rhythm and replicated as a template across additional programs.
Recognition
"
A huge success for Aero.
Leadership Recognition · FY2024–2025 · Exceeded Performance Rating
Adopted enterprise-wide as the template for program financial automation. Recognized across three consecutive annual reviews.
General AI RAG Development & Deployment — Pioneering AI in F&BO
The Opportunity
Finance & Business Operations lacked a validated framework for bringing LLM-based AI into regulated financial workflows — an early mover could define the enterprise AI playbook for the function.
The Solution
Designed and deployed a production RAG pipeline over a curated financial-data corpus. Established evaluation methodology, grounding checks, hallucination detection, and audit logging. Onboarded and aligned an external AI team to expand F&BO's AI footprint.
The Result
Improved financial data classification accuracy. External AI team now delivering major AI solutions across multiple F&BO customers.
Recognition
"
A significant milestone for enabling AI in F&BO.
Manager Recognition · FY2024 · Exceeded Performance Rating
Cited as pioneering work that set the AI adoption framework for the Finance & Business Operations function.
ContentShield Pro v3.0 — Forensic Watermarking & Attribution SaaS
The Opportunity
Digital content creators lacked a robust, production-ready tool for forensic attribution and unauthorized-distribution detection — an underserved niche at the intersection of AI-generated content and IP protection.
The Solution
Built full-stack SaaS independently: FastAPI backend, React frontend, Cloudflare Workers edge delivery, D1 databases, R2 object storage, OAuth 2.0 authentication, and forensic watermarking algorithms with tamper-detection and attribution logging.
The Result
Production-deployed SaaS with complete user auth, forensic attribution pipeline, and REST API — built solo, demonstrating full product ownership from architecture through deployment and real user delivery.
