AI & Data Advisory
I build data and AI platforms that have to be right.
Fifteen years turning enterprise data and AI from strategy into shipped, governed systems - across financial services, insurance, and life sciences. I also write about the part nobody warns you about: where agentic AI quietly breaks in production, and how to engineer around it.
Start here
I write long-form about what actually breaks when agentic AI goes to production - the failures that look like success, and the deterministic scaffolding that contains them. The Field Guide maps the whole series as a path to read; the essays go deep on one failure each; the reading list is the literature it all stands on.
About
I'm Tom Sullivan. Today I lead AI and data advisory work - building agentic systems for clients, from private-equity buy-side and sell-side diligence to enterprise data platforms. Recent work includes a schema-based document-processing framework that cut costs 80% across 1.1M documents a year on GCP Vertex, Document AI, Gemini, and Snowflake Cortex, and orchestrating ~960k production jobs at a 99.4% success rate. I was selected for Palantir's American Tech Fellowship.
Before that, at Discover Financial Services, I led the build of a high-performance ML platform - three teams, ~40 engineers - that accelerated model deployment by 60% and supported 50 models projected to drive $250MM in profit before tax. Earlier I led an automation organization that delivered $52MM in annual cost avoidance.
I hold an MS in Computational Analytics from Georgia Tech and a BS in Biophysics (cum laude) from Loyola University Chicago. My platform work earned the Gartner Eye of Innovation Award.
Track record
Elsewhere
LinkedIn · Timbre - a coaching system built on a deterministic rules engine and Claude · kew - an open-source governance layer for agent fleets (building toward launch).
The best way to reach me is LinkedIn - I'm always glad to compare notes with people building agentic systems in production.