AI engineer at HUMAIN, building multi-agent systems that automate enterprise workflows.
My research is on generative models for Saudi Sign Language. Next, I want to do a PhD on world models trained with reinforcement learning.
I'm an AI engineer at HUMAIN, where I work on HUMAIN Crew, a multi-agent system that automates enterprise agentic workflows.
I studied computer engineering at KFUPM with a research concentration in AI, spent an exchange term at Georgia Tech, and took advanced AI coursework at Oxford as one of 15 students selected from more than 7,000 applicants.
Before any of that I was on Saudi Arabia's mathematics olympiad team, and I still reach for a proof when a model misbehaves.
HUMAIN, Riyadh. 2026 to now
HUMAIN Crew: teams of agents that plan, call tools and hand work to each other to run enterprise workflows end to end.
SDAIA–KFUPM Joint Research Center for AI. 2024, and 2025 to 2026
Sign language video generation with Dr. Motaz Alfarraj: diffusion transformers conditioned on retrieved VQ-VAE pose tokens, trained on SLURM clusters.
Before that, with Dr. Abduljabbar Siddiqui: adversarial patches that still work in the rain, with a data pipeline that generates training data on the fly.
Baker Hughes, Houston. Summer 2025
Recovered low-frequency signals from distributed acoustic sensing with compressed sensing, and built abnormality detection models for embedded systems. Advised by Dr. Peyman Moradi.
OLIVES Lab, Georgia Tech. Fall 2024
Cleaned, documented and published terabyte-scale seismic and ophthalmology datasets on Hugging Face, with Prof. Ghassan AlRegib's lab.
My undergraduate thesis. A diffusion transformer generates Saudi Sign Language video from several conditioning signals at once.
It grew out of my proposal, SSLVG, a text-to-gloss-to-video pipeline: each sign's pose is retrieved as a VQ-VAE latent, a transformer models the transitions between signs, and the result conditions a DiT. Evaluated on Isharah-1000 and KArSL.
I want to do a PhD on world models: learning predictive representations of the world in the spirit of JEPA, and training them with reinforcement learning.
The architecture from Attention Is All You Need, implemented from scratch.
Games, shows and films I keep going back to, and five photographs of my own, all rebuilt as brick mosaics.
Minecraft, Counter-Strike, Half-Life 1, 2 and Alyx, Portal, Terraria, Cyberpunk 2077
One Piece, Attack on Titan, Spy×Family, Hunter×Hunter, Naruto
The Dark Knight, Interstellar, The Martian, The Odyssey, Avengers: Endgame
Mine, rebuilt in bricks like the rest. Tap one to see the original.
Middle East
Europe
Asia Americas
Bouldering, climbing and hiking, usually with a camera somewhere in the bag.
Built from bricks with three.js. Minifigure parts by BentheCreator and Lego Batman by Neut2000 (CC BY 4.0, Sketchfab), with hair and face prints from the LDraw parts library (CC BY 4.0). Minecraft skins from SkinsMC. Gallery mosaics are fan-made from each title's key art, which belongs to its owners.
Sorting bricks…
