Computer science (B.S., GPA 3.97/4.00, expected January 2027), based in Niamey, Niger. Software developer at Gabera Software Solutions.
I study how security detectors are evaluated under real operating constraints: a limited number of alerts per day, imperfect labels, and data that cannot always be shared. My main project is AuthBench, an open evaluation harness for authentication anomaly detectors, built as sole author on the LANL data. On 369 million logon events and 162 red-team campaigns, none of ten detectors finds any campaign within 10 alerts per day. The preprint is not yet peer reviewed (v2, Zenodo).
Research interests: reproducible evaluation of authentication and intrusion-detection systems, imperfect ground truth, privacy-preserving collaboration between organizations.
Looking for: funded graduate studies in computer science (master's or PhD, security and privacy), starting in 2027.
Also built: AuthGuard (authentication system from scratch, ASP.NET Core and Angular, 137 tests) and Aegis-ID (read-only Microsoft Entra ID posture auditor, 15 IAM controls, 94 tests).
