
Steve Lattanzio
I’m an engineer by training and by nature. I work at the intersection of machine learning, forecasting, and risk—as a Principal Machine Learning Engineer at MetaMetrics, working in psychometrics and education technology, and as Chief Innovation Officer at Weaver Mine Capital Management, focused on quantitative finance.
I tend to think in systems and probabilities: building models from first principles, testing assumptions against evidence, and trying to understand what is signal, what is noise, and what remains genuinely uncertain. Much of my work centers on predictive modeling and the design of quantitative systems for making better decisions under uncertainty.
This site is a working portfolio and thinking space for ideas and projects that interest me—often involving uncertainty, forecasting, truth-seeking, and the challenge of presenting complex information with maximum clarity and minimal distortion.
It is, more or less, my form of doodling. Though I am known to doodle too.
Education
Duke UniversityDuke Football • Pratt Fellow
Selected Projects
The projects below are attempts to make uncertainty legible.
Sports Analytics
A scientific framework for modeling game outcomes and team strength. Produces live rankings, spreads, total scores, and market comparisons.
College Scorecard
Higher education topological explorer mapping 1,776 colleges across 14 years of Scorecard metrics using 100 neural initializations.
College Atlas →Weather Engine
Minimalist weather application conveying probabilistic forecasts for local temperature, precipitation, and conditions with minimal noise.
Forecast →Truth Engine
A carefully-crafted LLM answering system compressing complex topics into dense, direct, and truthful answers with minimal distortion.
Query →Market Dashboard
A Tufte-inspired, ultra-minimalist intraday portfolio tracker visualizing real-time performance and weighted aggregation without chart junk.
View →