ML Engineer & AI Researcher
I build production ML systems, from classical ML and deep learning to LLM agents,
and research how to adapt LLMs to Arabic.
| Production ML | Generative AI | Research |
|---|---|---|
| Classical ML & ensembles (XGBoost, scikit-learn) Deep learning (PyTorch, TensorFlow) End-to-end MLOps on AWS (SageMaker, MLflow, Airflow, Docker) |
Agentic LLM workflows (LangGraph, LangChain) RAG assistants Voice AI LoRA / QLoRA fine-tuning (PEFT, DeepSpeed) |
Adapting LLMs to Arabic Parameter-efficient fine-tuning KAN / RBF feed-forward layers Arabic NLP & healthcare AI |
9 research publications · 570+ citations · Research at the University of Sharjah
Teaching: Learn PyTorch in 5 Projects on freeCodeCamp · Courses on Udemy
| Project | What it is |
|---|---|
| Arabic Medical QA: LoRA vs Full Fine-Tuning | 4 LLMs, 22 conditions, two-stage domain → task adaptation on multi-GPU · code · preprint |
| Text-to-SQL with Qwen2 | Qwen2-Instruct fine-tuned with TRL SFT · model on the Hub |
| Podcast Summarizer Agent | LangGraph agent that finds, transcribes and summarizes podcast episodes on its own · RAG, MongoDB |