Ishaku Gayus B.
Security & Systems Engineer · AI Safeguards and Evaluation
Summary
Security-focused systems engineer with 15+ years building and defending the networks and software of Nigeria’s national space agency (NASRDA). I wrote my organisation’s first cybersecurity policy, designed network security controls and audits, and led an automated monitoring system that predicts infrastructure failures before they happen (over NGN 100M saved a year). I now build AI safety evaluations: Salecta tests whether frontier models keep their safety behaviour in Nigerian Pidgin. I want to bring this defender’s mindset to AI safeguards full-time: safety evaluations, adversarial robustness and automated risk detection.
AI safety work
- Open evaluation of whether LLM safety behaviour transfers from English to Nigerian Pidgin, measuring both over-refusal of harmless requests and under-refusal of harmful ones.
- 100 matched prompts (60 harmless, 40 harmful) in three language versions, checked by native speakers; built in Python on the UK AISI Inspect framework with inter-annotator agreement analysis.
- Public code, data and an interactive “you be the safety filter” game: github.com/Lord-Shax/Salecta.
Experience
- Lead the ICT and software development divisions; set technical direction and mentor 25+ engineers and developers.
- Led the design and build of an IoT monitoring system that watches IT infrastructure continuously and uses predictive analytics to flag failures before they happen, saving over NGN 100M a year.
- Built an AI-powered IT support assistant (Python, FastAPI) that automates support workflows, and a machine-learning tool that surfaces recurring problems in user feedback for management action.
- Keep organisation-wide infrastructure, from servers to workstations, secure and available; build internal platforms, including an electronic document management system (in progress).
- Report technical metrics and project status to top management through Power BI dashboards; run staff training on emerging technologies, including AI.
- Wrote and implemented the Centre’s first cybersecurity policy; built and maintained its high-availability network infrastructure.
- Designed network security measures and ran regular audits against industry standards; deployed secure LANs and WANs.
Skills
- Security and detection
- Network security design, audits, cybersecurity policy, continuous monitoring, predictive early warning, automated alerting
- AI/ML and software
- Python, FastAPI, Inspect (LLM evaluations), machine learning for text, rapid prototyping
- Leadership
- Leading technical divisions, PMP-certified project delivery, mentoring, Power BI reporting, explaining risk to decision-makers
Education
University of Nigeria, Nsukka · 2017
Federal University of Technology, Yola · 2009
Certifications
- Project Management Professional (PMP)
- Cisco Network Security
- Policy for Science, Technology and Innovation · Harvard University
- Creative Problem-Solving · Imperial College London
Talks and service
- “Artificial Intelligence Ethics and Implications for Developing Societies” (talk)
- “Role of Information Technologists and Engineers in the Age of Artificial Intelligence” (talk)
- ITU WP 7B (WRC-27), ECOWAS region: reporting
- Member: IEEE, PMI, COREN, NSE