policy-evaluation

StatsPAI is the first Agent-native Python library for causal inference and applied econometrics — unified API, broad cross-method coverage, structured result objects, machine-readable schemas, Skills, an MCP server, and R/Stata parity validation.

Real randomized-trial causal analysis, honest uplift modelling, policy evaluation, and production experimentation design.

Quantile treatment effects in difference-in-differences for Stata (Callaway-Li 2019, copula stability): QTT with bootstrap CIs, uniform bands, covariates, and staggered adoption.

Stata package for difference-in-differences with a continuous treatment (Callaway–Goodman-Bacon–Sant'Anna 2024): B-spline dose-response ATT(d) & ACRT(d), uniform confidence bands, staggered adoption.

Review Physical AI with original evidence: compare VLM explanations against recorded robot outcomes, replay SmolVLA failures, and inspect Newton/Genesis simulations offline.

Research preview: vendor-neutral reference runtime and CLI for the Judgment Pack Specification (JPS).

End-to-End Python implementation of Rios, Barba & Gianmoena's (2026) Sender-Side sanctions incidence engine. Implements a two-wave CCE imputation counterfactual measuring the price and output costs borne by countries sanctioning Russia, with incremental vs level ATTs, bootstrap and placebo inference, and Russian-energy exposure gradients.

PRSM: Participatory System Mapping

Who should get the marketing e-mail? On a 64k-customer randomised test: everyone. Targeting models only looked better when scored on their own training data.

Anchore container analysis and scan provided as a GitHub Action

Mathematical study notes on Markov decision processes and dynamic programming, adapted from Puterman. Covers finite- and infinite-horizon MDPs, Bellman equations, policy evaluation, value iteration, policy improvement, and convergence proofs using Banach’s fixed-point theorem.

Research-grade epidemiological modeling, spatial metapopulation simulation, policy optimization, and Bayesian empirical Rt estimation engine in Python (FastAPI) and React 19.

A small, zero-dependency, async-native rule-evaluation engine — structural typing, real short-circuiting, and a verdict that always explains itself.

Stata package for difference-in-discontinuities (Picchetti–Pinto–Shinoki 2026): local polynomial estimation when a confounder sits at the same cutoff, plus two validity tests and partial-identification bounds.

Contract-gated safety layer for tool-using voice agents (EMNLP2026)

Statistical Evaluation Methods

You have a robot. orbit tells you what to do next, and how you will know it worked. Zero-dependency CLI for LeRobot datasets and evaluations: what the recording got wrong, what it never covered, whether a change survived the noise.

Audit recorded agent actions against approvals, revocations, and resource changes with evidence-linked policy decisions.

Contextual off-policy evaluation for calibrated local policy improvement and selective deployment.

Supply-chain simulation and RL policy evaluation: matched comparators, held-out tests and documented limits