About
Mission
I build intelligent systems — production-grade, rigorously evaluated, and designed to operate where the cost of failure is real
My work sits at the intersection of AI engineering and complex, interdependent systems: agentic pipelines that reason and act autonomously, deep learning systems deployed at scale on cloud infrastructure, retrieval architectures that ground language models in structured knowledge, and ML frameworks that quantify and predict cascading risk across interconnected networks. The domains vary — climate, infrastructure, logistics and supply chains, advanced air mobility, and project and asset management — but the standard doesn't: the system has to work, and it has to be trustworthy.
What drives that standard is a conviction I've held since my earliest research: intelligent AI is only useful if it's reliable. A model that performs in notebooks but fails in deployment isn't a solution. An agent that acts autonomously but can't be monitored or corrected isn't safe. The gap between impressive and dependable is where I do my best work.
Who I Am
Machine Learning Scientist and AI Researcher — one of the rarer hybrids who can take a system from research hypothesis to deployed production pipeline and back again
I design and build agentic AI systems, production deep learning pipelines, RAG architectures, and MLOps infrastructure — machine learning, sharpened by optimization and network theory, applied to risk and resilience in complex, interdependent systems. I also publish peer-reviewed research, hold Canada's most prestigious doctoral fellowship, and maintain open-source ML tools with thousands of real-world users.
My path has been anything but linear, and I think that's my greatest asset.
“
The universe is under no obligation to make sense to you.
— Neil deGrasse Tyson
Complex systems rarely do. And yet, understanding them — and building intelligence that navigates their uncertainty to turn risk into resilience — has been the defining pursuit of my career.
The Journey
A non-linear path that became the greatest asset
Egypt · 2011 — 2012
Before the vocabulary existed — an adaptive system, built at fourteen
Long before adaptive voice assistants were everywhere, I built one. At fourteen, I designed I system — a speech interface that personalized itself to each user. Where the speech-to-text engines of the day were rigid and hard-coded, I system gave every user their own profile, learned their individual speech patterns over time through a reinforcement-style feedback loop, and stored per-user corrections in a SQL database so it could recognize and act on spoken commands that were unique to them. It won at Intel ISEF Egypt in 2011, and a year later, at fifteen, I followed it with TRON OS. The core instinct behind both — systems that adapt to people rather than forcing people to adapt to them — is the same one that drives my work today.
Egypt · 2014 — 2020
Where it took root — valedictorian, highest honors, and a full tuition scholarship
That instinct — a fascination with how things work, and more importantly how they fail — set the direction for my undergraduate years. At the German University in Cairo, I threw myself into that pursuit fully, graduating as class valedictorian with highest honors, supported by a full tuition scholarship and multiple academic excellence awards.
Germany · 2018
The turning point — computational modeling, algorithmic thinking, and a shift in everything
The turn came during my six-month bachelor's thesis placement at the University of Stuttgart in Germany. It was my first serious encounter with computational modeling and algorithmic thinking — building software to simulate the behavior of complex systems. What I discovered there shifted everything: the most interesting questions weren't about individual components. They were about uncertainty, interdependence, and prediction. The instinct I'd had at fourteen finally had a discipline behind it.
Canada · 2020 — 2025
Vanier Scholar, PhD, and a full commitment to applied AI as a discipline
I moved to Canada to pursue a PhD at McMaster University, awarded the Vanier Canada Graduate Scholarship — one of the country's most prestigious doctoral fellowships. What began as research into ML-driven risk modeling evolved into something I hadn't anticipated: a full commitment to applied AI as a discipline in its own right. During my doctorate I stopped thinking of ML as a tool applied to other fields' problems and started thinking of it as the core problem — how do you build systems that learn reliably, generalize under distribution shift, scale to production, and remain interpretable when stakes are high? I built production TFX pipelines on GCP, developed graph-based ML systems for modeling cascading failures across interdependent networks, published 7+ peer-reviewed papers, and consolidated the domain that unifies my work — risk and resilience in complex systems. The most interesting unsolved problems in AI, I came to understand, aren't about model accuracy — they're about reliability, oversight, and what happens when systems operate beyond the training distribution.
2025 & Beyond · Present
Canada's Governor General's Gold Medal — and the work that matters most
In 2025, I completed my PhD and was honored with Canada's Governor General's Academic Gold Medal — the highest academic distinction awarded to a graduate student in the country. It was a milestone, but not a destination. Since then I've gone deeper into the problems that matter most to me: building agentic systems that can plan and act across long horizons, designing retrieval architectures that ground AI responses in verifiable knowledge, and building the monitoring and observability infrastructure that makes deployed AI systems trustworthy at scale — across climate, infrastructure, logistics and supply chains, advanced air mobility, and project and asset management. The technical frontier has never been more interesting — or the stakes higher. That's exactly where I want to be.
What Drives Me
Four convictions that show up in everything I build
Rigor
The work has to be done right
I hold my work to the highest standards — whether it's a published paper, a deployed ML pipeline, or a mentorship conversation. Good intentions are not enough. The work has to be right.
Impact
Research that moves the world
I build things that can be deployed, tested, and improved in practice — agentic pipelines running in production, open-source tools actively used by ML practitioners, and systems making real decisions at scale. The measure of good AI engineering isn't a benchmark number. It's whether the system holds up when it's actually used.
Ambition
Hard problems deserve bold thinking
The problems I care about — systemic risk and resilience, climate, intelligent logistics and supply chains, and safety in infrastructure delivery — sit at the intersection of multiple disciplines. They deserve researchers and engineers willing to think boldly, work across boundaries, and refuse easy answers.
Reliability
Can the investigated system be trusted?
The most important question in AI right now isn't whether a system can perform — it's whether it can be trusted. Trusted to behave consistently. Trusted to fail gracefully. Trusted to remain under meaningful human oversight as it scales. I think deeply about these questions in everything I build — not as constraints on what's possible, but as the actual engineering problem worth solving.
Beyond the Research
Mentorship, open science, and building things worth sharing
I am a committed mentor. Several of the students I've worked with have gone on to graduate programs and research careers of their own, and I consider that among my most meaningful contributions.
Teaching production ML beyond the classroom
My online course, Production-Grade Deep Learning Pipelines: TFX & Vertex AI, takes learners from notebook to a deployed, self-retraining pipeline on Google Cloud — eight hours of hands-on material with companion code on GitHub.
Explore the course →Open-source tools with real-world impact
Three Python packages published on PyPI — downloaded more than 4,000 times by ML practitioners worldwide. Built to fill genuine gaps in the ecosystem: metaheuristic hyperparameter tuning, multi-output TFX evaluation, and geospatial-to-graph conversion for spatial ML workflows.
Agentic AI systems — publicly documented and available
Autonomous research pipelines, multi-agent coordination frameworks, and RAG-grounded knowledge systems — all documented on GitHub. Built to be studied, extended, and used — not just demonstrated.
Writing about the technical problems worth thinking about
I write about the problems I find genuinely interesting — because the best ideas get better when they're exposed to scrutiny. If you want to go deeper, the projects speak for themselves.
Let's build something that matters
I'm open to roles and collaborations across applied AI/ML and research — spanning risk, resilience, and the complex systems where they matter — along with any technical conversation worth having.