About.

Most of my professional life sits between language and machine. I evaluate multilingual outputs at Innodata, focusing on linguistic fidelity in Semitic languages and cultural alignment across AI systems. The work is detailed, often unglamorous, and consequential in ways the LinkedIn version rarely captures.

Alongside that I'm finishing an MA in Cyber & Politics at Tel Aviv University — studying the political dimensions of digital infrastructure and how AI governance actually works (or doesn't) at the level of states, communities, and the people nearest the failure modes.

I've published a thriller novel (Mortal Past), two poetry collections (Crescent Moons and Nonperson), and one short story. A sequel novel (Deadly Sand) is drafting; a novella (My Name is Runner) is expanding into a full novel.

I also write a newsletter called The Loop — roughly monthly, on systems design, AI safety, and cyber politics. Essays live here first; LinkedIn is the republication. About 385 readers across both.

2018–22Masooda

An early-stage startup I co-founded. Wound down before scale.

2023→Itaha إتاحة

A non-profit advancing digital accessibility in the Arab community in Israel. Mostly advocacy + design education.

2025 (paused)Mawrid Research مورد

A small editorial & research studio for academic work in three languages. Currently closed to new projects while my attention is elsewhere; existing engagements continue.

2026→Folio

An integrity-first research workspace for academics — usefolio.co. Discovery, cited AI, surveyor, integrity tools. Publicly available with a free tier.

2026→Thagrah ثغرة

An open evaluation harness for adversarial prompts in Arabic — surfaces the safety gaps multilingual AI inherits from English-only alignment work. Currently building.

Originally from Baqa al-Gharbiyya, a Palestinian-Arab town in central Israel. Currently in Haifa — the city I keep coming back to, partly because it's where the work is and partly because of the light off the bay in late afternoons.

• · That multilingual AI safety is mostly an English conversation. The languages spoken by people most exposed to bad outcomes are the ones with the worst evaluations.
• · That accessibility is a baseline, not a flourish — the test is whether your work works for the people you haven't designed for.
• · That the politics of digital infrastructure are inseparable from the politics of the place it lives in — cleanly-engineered systems still leave their fingerprints on the communities they touch.
• · That fiction is a way of thinking that doesn't reduce.