Data & interface engineering / 2026
Market research workspace
Probabilities, with their sources attached.
The starting question
What happens when the sources disagree?
A research workspace for comparing prediction-market signals, inspecting changes over time, and reading the context around them. The interface keeps sources and uncertainty close to the numbers.
Private project. Working name and source kept private.
Python / Flask / React / TypeScript / SQLite
Concept study, not a product screenshot.
Inside the work
The details behind the interfaceDisagreement is information
Different sources do not always tell the same story. The workspace keeps their differences visible rather than smoothing everything into a single reassuring number.
History has to exist first
Movement is calculated from captured observations. When the history is too short, the interface says so. Missing evidence should be understandable, rather than quietly replaced with a zero.
From data to a usable view
A Python backend feeds a React research interface with charts, filters, and contextual reports. The design problem is helping someone inspect the evidence without losing sight of its limits.
The thread that carries through
Uncertainty belongs in the interface. A missing observation and a measured zero should never look like the same thing.