My path into AI runs through cybersecurity, not a straight computer science track. That shows up in how I build: I care more about systems holding up under real conditions than looking good in a demo.
My flagship project is a phishing detection platform running two fine-tuned RoBERTa classifiers, one for email and one for URLs, at 94% and 96% accuracy. I built it as a full FastAPI backend with JWT authentication, rate limiting, and Docker deployment, plus a Streamlit frontend for single and batch checks. The pipeline itself extracts URLs, routes by phishing precedence, and explains its own decisions through saliency highlighting, so it's not a black box.
At LASUSTECH, I lead practical sessions for 50+ students in cybersecurity, hardware, and systems management. I redesigned the lab and mentoring structure there, and proficiency scores rose by more than 25%.
My goal now is to become a professional AI engineer who builds production-grade RAG and agentic systems ones that act rather than just predict, with the same discipline around security and explainability I bring to my work today. Leadership that stays values-driven under pressure matters to me too, and it's a big reason mentoring stays part of my work alongside the engineering.
