AI Engineer. I build end-to-end ML and LLM-powered systems β from RAG pipelines and computer vision APIs to MLOps tooling. Currently contributing to open source at NumFOCUS/ArviZ and exploring scalable AI deployment.
- π B.Eng. Communication & Information Engineering β Helwan University (2022β2027)
- π Currently building: ArabiQ β Arabic Document Intelligence with hybrid RAG
- π± Expanding into: MLOps, CI/CD for ML, scalable LLM serving
- π ECPC 2025 Participant β solved 4 problems
Languages
ML / AI
DevOps & Tools
πΈπ¦ ArabiQ β Arabic Q&A System
Python Β· LangChain Β· FastAPI Β· RAG Β· BM25 Β· Dense Retrieval
Arabic-aware RAG system with sentence-level chunking, hybrid retrieval (dense + BM25), and Ξ±-tuned re-ranking. Improved Precision@3 from 62% β 81% on a custom 250-pair Arabic QA dataset. Deployed as a FastAPI service with sub-1.5s end-to-end latency.
Python Β· YOLOv8 Β· FastAPI Β· Streamlit Β· Docker
YOLOv8 model trained on 5K+ annotated food images achieving 0.82 mAP@50. Dockerized FastAPI + Streamlit pipeline with 30% latency reduction for real-time inference.
NumFOCUS / ArviZ β Active contributor since Apr 2026
Standardizing DevOps tooling (tox, mypy, pytest) across modules; building CI workflows for upstream scientific Python testing.
"Build things that work. Ship things that matter."