Peer-reviewable work, written for people who make decisions.

A diagnostic layer on top of existing forecasting stacks: regime-aware evaluation, provably leak-safe backtesting, and an economic decision layer that prices forecasts in currency instead of abstract error.

Innovation-diffusion forecasting for cold-start adoption curves: Bass models, a stochastic differential equation on the logit scale, and era-aware Bayesian priors — tested on 17 historical technologies.

Event extraction, graph neural networks and causal inference for resilience insight — alongside a deliberately skeptical test of whether feasibility can be mathematically certified at all.

A bio-inspired reliability optimizer for planning under uncertainty — selective risk avoidance, detour generation, outcome memory and adaptive exploration — now running behind domain-neutral contracts on real supply-chain data.

Lightweight extraction of causal event graphs from unstructured text — financial, political and general news — with a stable schema, provenance on every edge, and optional market-impact annotation.

Five neural time-series models blended through a regime-aware, online-adaptive ensemble for Brent crude — validated through a live +103% geopolitical price shock.

A Sharpe-ratio optimization framework that turns ML forecast output into disciplined, risk-aware portfolio allocation.

A natural-language-first tool that helps warehouse and category managers understand their inventory, forecast what is coming, run what-if scenarios and get explainable, auto-selected optimization recommendations.

Transformer surrogates in the prior-fitted-network style for expensive, constrained engineering optimization — packaged as a reusable tool with a license-clean benchmark suite and an interactive comparison of search strategies.

A lightweight, scikit-learn-compatible package for individual causal treatment effects (CATE) in campaigns, A/B tests and feature rollouts — one API across meta-learners plus a reproducible multi-dataset benchmark of when each one works.

An exploratory analysis of more than 7,500 pirate attacks on cargo ships, combined with World Bank and FRED country indicators — trends, hotspots, seasonality and significance-tested correlations.