Innovative AI Engineer and Data Scientist with a B.Sc. in Artificial Intelligence (Honors) and hands-on experience in Distributed Computing and Generative AI. Proficient in the full ML lifecycle—from fine-tuning LLMs (Llama 3, QLoRA) and implementing RAG pipelines to deploying production-ready applications using Django and React.

• 📋 Resume: Check it out here
• 💬 Ask me about: Generative AI, LLMs, RAG Pipelines, Machine Learning, Deep Learning, Data Science
• 🧐 Interested in: NLP, Computer Vision, Distributed AI, MLOps
• 📫 Reach me at: dev.mahmoudrefaey@gmail.com | +20 1026295189
• 📍 Location: Cairo, Egypt

• 
• 
• 
• 📚 Key Courses: Machine Learning, Deep Learning, NLP, Computer Vision, Reinforcement Learning, Database Systems, Data Structures & Algorithms, OOP
• 🧪 Graduation Project: AI-Powered Churn Prediction Platform for Telecom Companies

Beetleware · Remote Internship
Aug 2025 – Jan 2026

• Built distributed AI models using the Ray framework and contributed to open-source projects.
• Developed and deployed full-stack SaaS features with modern frameworks, APIs, and CI/CD.
• Collaborated on real company projects in Agile teams, delivering production-ready solutions.

Digital Egypt Pioneers Initiative (DEPI) · Hybrid Traineeship
Oct 2024 – May 2025

• Built and deployed machine learning models using Python, Pandas, scikit-learn, and TensorFlow.
• Collaborated in team projects simulating industry challenges, including data preprocessing and model evaluation.
• Mentored and supported peers, helping them master core concepts in machine learning and data analysis.

"Customer Churn in Telecom: Predictive Modeling for Enhanced Retention Strategies" | Egyptian Russian University

• Engineered a stacking ensemble (XGB-LGBM-RF) with 91% CV accuracy, deployed via a Django-React full-stack predictive platform.

"A Study of Generative Approaches for Balancing Imbalanced Data: SMOTE, GANs, and LLMs" | ResearchGate

• Comparative study of SMOTE, GANs (CTGAN/TVAE), and LLMs for fraud detection, concluding traditional sampling offers superior stability over generative models.

• Proven ability to analyze complex problems and devise data-driven solutions.
• Highly adept at evaluating models and making decisions based on analytical insights.

• Experience working in interdisciplinary teams to develop AI solutions.
• Strong communication skills with both technical and non-technical stakeholders.

• Passionate about solving unconventional challenges through innovative use of AI and data.

• Ability to explain technical concepts clearly to non-technical stakeholders.

• Led group projects and mentored junior team members in machine learning methodologies.

• Efficiently managed overlapping academic and internship responsibilities while delivering high-quality work.

• Rigorous in model tuning, ensuring precision and accuracy in results.

• Consistently driven to improve technical skills and apply them in real-world scenarios.

• 🇪🇬 Arabic: Native
• 🇬🇧 English: Fluent

AI Engineer – Full-Stack Developer | React 19, TypeScript, Vite, Python, LLMs

• Developed an AI-powered document management ecosystem featuring intelligent document, data, and PDF editors with automated content generation, custom file formats (.nd, .np, .ndf), and robust file-storage architecture.
• Enhanced communication by connecting AI services with real-time chat (Socket.io) and internal email modules—providing smart, context-aware collaboration.

AI Engineer | Python, RAG, TinyLlama, DialoGPT, Phi-2

• Engineered a modular RAG pipeline using Streamlit and FAISS for real-time semantic search and Q&A over multiple PDF documents.
• Integrated local, open-source LLMs (TinyLlama, Phi-2) and Sentence-Transformers to enable private, cost-effective document intelligence.

Data Scientist – Backend – Team Leader | Python, Django, Machine Learning

• Developed a high-performance stacking ensemble (XGBoost, LightGBM, RF) achieving 91% CV accuracy and 84.5% ROC-AUC for telecom retention.
• Architected a multi-tenant Django REST backend with JWT authentication and real-time prediction APIs featuring probability scoring.
• Led the end-to-end development life cycle as Team Leader, from technical documentation to full-stack deployment.

Project	Links
🦙 Fine-Tuning LLaMA3 for Coding Tasks
⚡ Energy Consumption Forecasting
🍕 Food Classification ViT Model
🛰️ Land Cover Classification with ResNet50 (EuroSAT)