Data Scientist · ML Engineer
Addison Sans
Available for contract work — remote or hybrid
I build AI and data pipelines that turn messy, high-stakes data into clean, usable systems. Background spanning neuroscience research, industry data science, and government regulatory data.
Comfortable in environments where the data is messy, the requirements are unclear, and the output still has to be right.
Selected projects
All projects →- The Founding Group Chat 2026
A shareable RAG web app where the Founding Fathers (plus Abigail Adams) debate modern 2026 topics in a live group chat, each reply grounded in their real writings.
Python · FastAPI · Anthropic Claude · LanceDB · fastembed · Supabase/Postgres · Fly.io
- CILB Document Pipeline 2026
AI automation system built for the Florida Construction Industry Licensing Board, replacing a manual paper-based document review process used in official licensing proceedings.
Python · OCR · LLM extraction · Vision APIs · SQLite
- Bakery Daily-Delivery Forecasting 2024
Forecasting out-of-stock events for a California bakery's daily-delivery model — Python analysis behind Tableau dashboards the operations team actually used. Engagement continued under contract after I rolled off.
Python · Pandas · Tableau · Time-series forecasting
- The Tale of Two Collapses 2026
A win-probability study of two of the biggest blown leads in sports — the Spurs' 29-point 2026 Finals collapse vs. the Falcons' 28-3 — plus an NBA contract-value analysis, all from live ESPN and Basketball-Reference data.
Python · pandas · matplotlib · seaborn · ESPN API · Basketball-Reference
Publications
Research →Co-authored peer-reviewed work from my time in neuroscience research.
-
Forthcoming paper (co-authored with David Rosenthal and Brian Odegaard) — pending submission.