Callum Doty
Product Developer & Designer
Developer and analytical problem-solver based in Kansas City, focused on building systems that turn complex data into actionable decisions. My work spans statistical modeling, software development, and product design.
Focus and Interests
I enjoy tackling problems where rigorous analysis and well-designed software can meaningfully improve how teams make decisions. I’m particularly drawn to Bayesian reasoning, systems internals, and AI.
Selected Projects
- Campaign Allocation
Where should a party committee spend its money across House races? A calibrated forecast of every race, an explicit assumption about what party money does, and a test of DCCC's real spending against the best use of the same budget.
- Kalshi Weather Derivatives
A quantitative trading system for Kalshi's daily high-temperature markets: probabilistic forecasts, a market-facing scoring benchmark, and serverless execution live with real capital across 7 U.S. metros.
- Axiom Catalog
An AI document intelligence platform for political campaign materials, using Claude, hybrid semantic search, and automated Dropbox ingestion to make 9,800+ past pieces searchable in plain language.
Recent Articles
- The Persuasion Ceiling
Debugging a House-race spending optimizer that projected an implausible 13-seat gain, tracing the bug to an unbounded marginal-return term, and fixing it with an endogenous persuasion ceiling.
- Political Theta
Building a House-race allocation model on DCCC and FEC data and deriving Θ, a Bellman-equation measure of when a campaign should spend versus wait. A later rebuild didn't support the misallocation it seemed to find.
- Chess to Problems
Chess openings as a window into decision architecture: what repertoire choices reveal about risk tolerance, complexity, and how I approach problems beyond the board.
- Why My Model Trained Differently on a CPU and a GPU
Why identical Temporal Fusion Transformer training runs diverge between CPU and GPU — floating-point rounding, reduction order, and the hidden nondeterminism of deep learning.
- DUI: Drinking, Driving, and Bayesian Analysis
What if I told you that based on a positive breathalyzer test, the probability that you were driving drunk is less than 2%?