I'm a Data Science and AI senior at Zewail City (graduating June 2026), based in Cairo. Most of what's in this profile came out of trying to build full systems instead of single notebooks: a graduation thesis with a mobile app, an RL agent, and an LLM coaching layer all talking to each other; a couple of RAG pipelines for medical and document QA; an ML CI/CD setup that gates deployment on actual accuracy thresholds.
I like the parts of ML that don't show up in tutorials: getting a model from a notebook into something a user can actually open, and making training pipelines that don't quietly break.
Currently: finishing my thesis, Captainy, an AI gym assistant that combines computer vision form-checking, a DQN-based workout planner, and a LangGraph coaching agent (led a 4-person team on it; code isn't public yet).
- cxr-rag-system — chest X-ray RAG system combining ColPali patch-level retrieval with MedGemma, benchmarked against a CLIP baseline across 50 test studies.
- DSAI-413-Multi-Modal-RAG-QA-System — vision-first RAG over PDFs (text, tables, charts) using ColQwen2.5 and Llama 4 Maverick, with a Streamlit demo.
- ml-cicd-gatekeeper — GitHub Actions pipeline that trains a model, logs to MLflow, and only deploys past an accuracy gate.
- Phonify — bilingual (English/Arabic) Flutter shopping app with full RTL support.
Python, PyTorch, TensorFlow, Scikit-learn, Transformers · Flutter/Dart · SQL, Power BI, Tableau