Mohamed Atr

I’m an AI Engineer and Security Researcher focused on architecting Autonomous Agents and kernel-level security frameworks. My work centers on turning advanced reasoning into functional, resilient execution—so systems don’t just “think,” they safely act.

I leverage LangGraph and LangChain with LLMs to build self-correcting workflows that can reason through multi-step tasks independently. I’m especially interested in bridging the gap between model capabilities and real operational reliability in security contexts.

In freelance remote work, I used prompt engineering and AI agents to architect and debug complex security frameworks, reducing development time by 50%. I also integrated backend automation in Python with AI-assisted pair programming to maintain high code quality and security.

My projects reflect this direction: PhantomForge/GhostThread, an AI-native endpoint detection and response platform designed for authorized security testing, and autonomous red-team assessment concepts aimed at reconnaissance, exploitation, and post-exploitation. I bring a security-first mindset—using self-healing concepts and rigorous execution to minimize false positives and maximize resilience.