Supplementary material for "Auditing Gender Disparities in an Academic Profile Retrieval Pipeline": scripts and aggregate results

Scripts and aggregate results accompanying the manuscript "Auditing Gender Disparities in an Academic Profile Retrieval Pipeline: Indexing Choices and the Evaluation of Preprocessing Mitigation". The package contains (i) a held-out comparison of five preprocessing fairness techniques (Holistic AI) over 30 stratified splits, (ii) an audit of the gender composition of retrieved researcher profiles for 220 research-area queries under three variants of the indexed text, and (iii) checks reproducing the original in-sample protocol. Profile-level data are not included; the result files contain aggregate statistics only and no author identifiers. The scripts were written and run with the assistance of Claude (Anthropic) and reviewed by the author.