“You love work, you don’t love me.”
- LiteralA complaint about work.
- Between the lines“I miss you.”
- FeelingLonely, and not sure she still comes first.
- FixClose the laptop. The bug will still be there tomorrow.
بسم الله الرحمن الرحيم
I take things apart until the signal shows: systems, data, and the sentence someone didn't quite say.
Hold still and the picture clears. Ironically, I used to measure that for a living.
Every fact has a layer underneath. Hover with the lantern, tap on a phone, or tab through.
Head of Data Science at Keak.
A small team that ships like a big one. I turn messy questions into experiments that can't lie, and I go through a lot of whiteboard markers. You can vibe-code a task, not a system.
PhD in Computer Science, University of Ottawa.
As a kid I wrote د. م. on my bedroom door: Dr. Eng., not the medical kind. It took three countries and a heroic number of rejections to make the door honest.
MSc: telling Liverpool fans from Chelsea fans, without labels.
It started as a different topic. Topics change. A method that generalizes doesn't care.
Left a well-paying bank job after six months.
Some things don't fit in a salary. Principles make up your spine. Uphold them and stand tall. Bend, and you bend.
Spent four years measuring how engaged people are, from brainwaves, eyes and gameplay.
Favourite finding: people read engagement off a face at 0.48 F1, about as well as a coin toss. What people do tells you more than what you are told.
Most cited paper: how people are offensive in Arabic on Twitter.
Rudeness turns out to be measurable. So is kindness; it just gets fewer citations. I'd still rather measure the noise than add to it.
Valedictorian, 3.97 GPA.
The GPA kept my tuition waived after year one. The rest came from $2-an-hour campus jobs, because I didn't want to take money from my dad.
750+ lessons to kids, retirees, and future bank executives.
Favourite student: a chemistry teacher who went from zero Python to his first data job. Learn to teach, and teach to learn.
Finished Skyrim, Elden Ring and Sekiro, several times.
That used to be how I rested. Now it's snow, mountains and the ocean. Games add noise. Nature takes it away.
Deep focus is my default state.
Until my daughter hugged me and cried, "You love work, you don't love me." Being there for the people I work for turns out to be good for the work too.
My one trick: take anything apart until its pieces make sense. A system, a dashboard, or something said at dinner.
“You love work, you don’t love me.”
“Can we make this dashboard prettier?”
“The model is 99% accurate.”
“AI is too expensive.”
From a small town in the Nile Delta to the coldest capital in the world. Scroll and each place comes into focus.
Where I was born and learned to ask why. Full marks in physics, strong opinions on everything. Some places you carry at a resolution no camera has.

Seventeen, a scholarship, and fifteen million lights. Computer science, taught by people who also made me read philosophy, sociology and history. It worked.

Three papers as an undergraduate on serious games for children, a research tour from Brussels to Ljubljana, and a new passport. Still the kid asking why.

My first international conference, as a volunteer. I flew home with a research collaboration and an idea, sentence embeddings plus a humble SVM, that later took first place in a shared task.

MSc done, first-author paper at ICWSM. Water this clear makes you suspicious of every blurry thing you were ever told.

My last winter in Türkiye. Two months later, a one-way flight to Ottawa and a PhD.

The coldest capital in the world. People tell me that like it's a complaint.

All that water, all that roar, and still a rainbow in the spray. Signal survives noise.

Twenty minutes from home, the world gets big and I get to be the right size.

I defended my PhD from here, remotely, with the Atlantic just out of frame. Four journals, one conference, and a committee that smiled.

Hiking in summer, snow in winter, data science in between. 8,770 km from Bilqas, before thirty, and somehow still the same kid.
My life embedded, projected and clustered with the same method as my stance-detection paper (UMAP, then HDBSCAN). Hover a point to read it.
A measurement without its uncertainty is just an opinion.
Google Scholar, November 2026. h-index 8.
العربية, English, Türkçe, et un peu de français.
Egypt, Türkiye, Canada.
Enjoyed: 5 ± 0.
Warak enab, p < 0.01. İskender and Nashville hot chicken still under review.
Instrument saturated. Results not reported.
Research that passed review, and code you can read.
Keak runs conversion experiments on websites by itself. I lead the data science behind it: generating variants, sequential statistics that call winners honestly, and feedback loops that make every test teach the next one.
Cheaper Inference is how we push the price of AI down: one API, hundreds of models, never above list price. Built by a team that makes hard things look easy.
Before: ML engineering, an SME-lending bank, a computer-vision system that graded pizza for 20+ restaurants, and four years of consulting and teaching.
Code
Tweets → sentence embeddings → UMAP → HDBSCAN. 90% precision, zero labels. The method behind the map above.
My BSc graduation project: a university's social network, crawled, clustered and told as a story over time.
Gene expression → metabolic flux → disease classification. Undergrad bioinformatics, with linear programs built on the fly.
Pooled vs separate vs hierarchical Bayes for blood pressure, in Julia. A lesson in honest uncertainty.
وَعَلَّمَكَ مَا لَمْ تَكُن تَعْلَمُ ۚ وَكَانَ فَضْلُ اللَّهِ عَلَيْكَ عَظِيمًا
“And He taught you that which you did not know. And ever has the favour of Allah upon you been great.”
Qur’an 4:113