I build software and have led IT teams — and I research why AI coding agents get rejected, and what it takes to make autonomous, LLM-based systems trustworthy teammates.
Each contribution carries its real state — merged into a venue, open for review, or a public draft.
Public data behind the research — how AI coding agents and humans contribute, at the scale of millions of pull requests.
Beyond the papers — data products, tools, and apps I’ve built and shipped.
Doctoral research at ÉTS Montréal, plus building, teaching, and leading IT at UCAS in Gaza.
From a diploma in Gaza to a doctorate in Montréal — a decade of computer science and software engineering.
Competitive, merit-based doctoral funding from the Fonds de recherche du Québec (FRQ), CAD $56,250 over three years.
International Conference on Business and Technology (ICBT Istanbul) — machine learning for education.
Supervised the student project that won 1st Place, Secure Digital Innovation 2026 (PCSAi) — bidirectional ASL↔text via deep learning, computer vision, and a real-time 3D avatar.
Peer review for the 22nd International Conference on Intelligent Environments.
AI coding agents now open pull requests, refactor code, and propose fixes — yet a large share of that work is quietly discarded. My research treats those rejections as data: what does the gap between what an agent produces and what a maintainer will accept tell us about building AI that developers can actually trust? I work empirically, grounded in real repositories and real review decisions, across AI for software engineering, automated program repair, and large language models.