LAB INDEX / SECONDARY WORK

Experiments kept in proportion.

Supporting systems, benchmarks, and delivery artifacts. The flagship work remains Calybris Core, ProofFrame, and ReproCut.

ForecastEdge ↗

Archived forecasts scored against the guess available without them. Python, Rust, and verification.

Criteo uplift benchmark ↗

Five causal learners evaluated on the real seven-million-row Criteo dataset.

Aegis ↗

A bounded workbench for the boundary between prediction and prescription.

Aera ↗

Rust and Polars experiments for deterministic financial analysis.