AI Engineer specializing in LLM systems, agentic architectures & computer vision. Building scalable, production-ready machine learning pipelines with real-world impact.
I'm an AI Engineer specializing in building scalable SaaS-based machine learning systems across Computer Vision, NLP, and Generative AI. Currently pursuing an M.Sc. in Computer Science (Artificial Intelligence specialization) at the Faculty of Computers and Artificial Intelligence, Cairo University, strengthening both theoretical foundations and advanced AI research knowledge.
Experienced in designing and deploying end-to-end AI pipelines, including data processing, model development, evaluation, and production deployment. I am proficient in Python and deep learning frameworks such as TensorFlow and PyTorch, and I build production-ready AI applications using FastAPI, Streamlit, and Docker.
I work extensively with LLMs and Generative AI frameworks including LangChain, Hugging Face, and Retrieval-Augmented Generation (RAG). I am also advancing in Agentic AI and AI Agents, including ReAct, LangGraph, multi-agent systems, Agent-to-Agent (A2A) communication, and Model Context Protocol (MCP) to develop autonomous and scalable AI solutions.
Additionally, I am expanding my expertise in AWS Cloud (including Amazon Bedrock and SageMaker) and Data Engineering, focusing on data pipelines, ETL processes, data modeling, and scalable data architectures to support robust, production-grade AI systems.
An agentic AI assistant built with Mistral LLM, LangGraph, and custom tool-orchestration. Integrated Tavily for web browsing and SendGrid for direct email actions. Deployed on Streamlit Cloud with adaptive memory and seamless UX designed to automate research and multi-step tasks.
Enterprise chatbot leveraging Retrieval-Augmented Generation to answer company-specific queries with accurate, grounded responses. Built for GlobalCorp's internal knowledge management and operational efficiency.
AI-powered financial analysis tool for GlobalCorp & Ollin Group. Parses bank statements, extracts key financial signals, and surfaces actionable insights with automated reporting capabilities.
Streamlit app performing customer segmentation via K-Means clustering. Users select features, determine optimal clusters via the Elbow method, and visualize distinct customer groups with autosummaries and profile tables.
XGBoost classifier predicting obesity level from age, gender, weight, and lifestyle habits. High-accuracy multi-class classification with real-time interactive predictions and visual model explanations via Streamlit.
ML model trained on clinical data — blood pressure, cholesterol, ejection fraction — to predict heart failure risk. Designed to support early diagnosis and preventative care through data-driven insights.
Automated attendance tracking via facial recognition using CNNs, OpenCV, and FaceNet. Real-time face detection and identification with database logging — eliminating proxy attendance entirely.
TensorFlow and LSTM-based sequence-to-sequence translation model. Encoder LSTM compresses input into a context vector; decoder LSTM generates target language word-by-word from bilingual training datasets.
Open to collaboration, project inquiries, and meaningful conversations.
Whether you have an idea or just want to connect — reach out.