Research curiosity.
Engineering instinct.

I work where learning meets deployment.

My research asks how we can make AI more capable within practical constraints. I connect Bayesian methods, model compression, and physics-guided learning with the engineering needed to run models on real devices.

At ZEISS, I work on machine learning for generative AI and computer vision, including the optimization of foundation models for edge accelerators. Alongside this, I conduct independent research with collaborators in the INSAIT and ETH Zurich research ecosystem.

My first-author work includes PhysDNet, Quant-Trim, and Shaving Weights with Occam’s Razor. These projects approach efficiency from different directions: the learning signal, numerical precision, and the model’s capacity to be pruned.

A perspective shaped by the full system.

I studied Robotics, Cognition, Intelligence at the Technical University of Munich, following a bachelor’s degree in Electrical Engineering & IT. A research stay at EPFL and experience at ETH Zurich, Intel, BMW, Huawei, CARIAD, and Infineon shaped how I approach problems across software, models, and hardware.

I enjoy turning research questions into working systems, then using what deployment reveals to ask better questions.

A little personality, too.

The curiosity I had as a child still guides me: the urge to ask why, explore unfamiliar ground, and let the world surprise me. Learning changes the questions I can ask as much as the answers I can give. Each new understanding opens another horizon, and I hope to keep that sense of possibility alive.

Moving between research and engineering keeps me attentive to both ideas and their consequences. I want to understand things deeply, question my own assumptions, and turn what I learn into something useful. Growth, for me, includes allowing experience to revise my thinking.

I believe in keeping moving, whatever the pace. Whether it feels like 1,000 km/h or 10 km/h, what matters is continuing forward with curiosity and purpose. Persistence also means being willing to adjust my course as I learn. I try to meet each day with a little more understanding, and to let that understanding shape how I live and what I build.

Research as a shared practice.

I review for NeurIPS, ICML, ICLR, AAMAS, and JAIR. I speak Arabic and French natively, and English and German fluently.