Beyond Model Audits: Multi-Factor Evaluation of A/B Configuration Integrity in AI Systems
Year: 2026
Type: article
Abstract: This ZIP provides a Python reference implementation supporting the manuscript “Beyond Model Audits: Multi-Factor Evaluation of A/B Configuration Integrity in AI Systems.” It includes flag-value and pairwise coverage calculations, segment parity and regression parity metrics, risk scoring, deployment-gate decision logic, and Coverage Equity Deficit analysis. The package also contains a trial-analysis utility, documentation, an input-data schema, and 35 passing tests covering selected calculations and software behavior. A clearly labeled synthetic dataset demonstrates the analysis workflow; it does not represent original experimental observations. The package supports inspection and reuse of the implemented methods rather than independent replication of the entire empirical study. The software is available under the Apache 2.0 license.
Author: Yeruva, Joseph Sukanth
Institution: Apple (United States)
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7236744387
