عبد الكريم الخطيب

Abdelkareem Elkhateb

I teach machines to understand Arabic.

300 million people speak Arabic, but most AI tools ignore them.

I’m changing that — building models that actually work for Arabic speakers, even on a cheap phone.

GitHub

Research

Hire Me

What I’ve Built

Zarra & Bojji

Arabic AI models that are 10x smaller than competitors but just as accurate. They run on phones, not just expensive servers.

Why it matters: A student in rural Egypt can now use Arabic AI without cloud costs.

كم كالوري

Ask “how many calories in koshari?” in Arabic and get real answers. Semantic search for nutrition.

Try it →

GPUVec

Find the cheapest GPU to train your model. Compares 50+ cloud providers in real-time.

Check prices →

Arabic GLiNER

Extract names, places, and organizations from Arabic text automatically.

GitHub →

Where I Work

xbites 2025 - now

At xbites, I developed the AI and search solutions for the Darin AI Smarted Real Estate System. My work included organizing and integrating data from over 100 different developers, designing a custom search engine, and building self-improving AI agents to address complex business challenges.

Result: Clients are now able to process documents 80% faster.

Hamza Salem Lab 2024 - now

I am part of a research group focused on advancing Arabic AI. My responsibilities involve building embedding models and contributing to academic publications. Our mission is to make cutting-edge Arabic AI widely accessible beyond large technology companies.

Result: 7 papers published, 43 citations, and our models are used internationally.

NAMMA Nov 2024 - Nov 2025

As a part-time NLP Engineer and open-source maintainer at NAMMA, I contribute to state-of-the-art Arabic large language models, embeddings, OCR, and ASR systems. I help train, evaluate, and release open models that power Arabic AI applications worldwide while advancing research through community-driven development.

Result: Multiple SOTA Arabic models released openly, used by thousands of developers and researchers across the Arab world and beyond.

Freelance 2021 - now

As a freelance developer, I deliver tailored solutions for clients, including search engines, APIs, machine learning pipelines, and web applications. I am a top-rated freelancer on Upwork with more than 20 successfully completed projects.

Result: Repeat clients from over 5 countries.

What I’m Working On Now

Better Arabic embeddings — Current models still struggle with dialects. Fixing that.
Arabic OCR — Reading old Arabic manuscripts and printed books automatically.
Multimodal Arabic AI — Models that understand both Arabic text and images together.

7 Papers

43 Citations

962 Papers Read

Work With Me

Need help with Arabic AI, ML systems, or web development?

Email Me Upwork

Updates

Get notified when I publish new research or projects.