I build production-grade AI systems powered by LLMs, focusing on scalable backend architectures, Retrieval-Augmented Generation (RAG), and real-world deployment.

Currently working as AI Engineer at Notyet Studio, where I design and deploy AI systems handling thousands of daily requests with high availability (99.5% uptime).

• RAG Systems: End-to-end pipelines (ingestion → chunking → embeddings → retrieval → generation)
• LLM Integration: OpenAI, Cohere, Google GenAI
• Prompt Engineering: Optimization, evaluation, and response control
• Semantic Search: Embeddings, hybrid retrieval (Qdrant + pgvector)
• LLM Orchestration: LangChain-based pipelines

• APIs: FastAPI (async), RESTful services
• Performance: Async processing, caching, latency optimization
• Architecture: Config-driven systems (.env-based), scalable design
• Concurrency: Handling parallel LLM requests efficiently

• Vector DBs: Qdrant, pgvector
• Databases: PostgreSQL, MongoDB
• Pipelines: PDF ingestion, parsing (PyMuPDF), data preprocessing

• Background Workers: Celery, Redis, RabbitMQ
• Deployment: Docker, Nginx, CI/CD
• Monitoring: Prometheus + Grafana (latency, throughput, health)

• Natural language data analysis — ask questions in English or Arabic, get charts and insights
• LLM pipeline: intent parsing → semantic validation → pandas execution
• Bilingual support (AR/EN) with RTL/LTR UI
• Fingerprint-based 24h trial with session isolation
• Fully containerized with Docker, Nginx, Prometheus, and Grafana

Backend • Frontend • Docker

• Built end-to-end RAG pipeline for document-based Q&A
• Designed hybrid retrieval system (Qdrant + pgvector)
• Implemented async APIs handling concurrent LLM requests
• Integrated multiple providers: OpenAI, Cohere, Google GenAI
• Added caching + background processing (Redis, Celery)
• Built observability stack using Prometheus & Grafana

• Custom CNN-based face detection pipeline
• 95%+ accuracy, real-time inference
• End-to-end ML lifecycle (data → training → deployment)

• Real-time object detection using YOLO
• 96% accuracy on video streams
• Optimized for real-time performance

• Medical imaging classification using VGG16
• Improved performance by +18% over baseline
• Applied transfer learning & augmentation

• OpenCV, CNN, Image Preprocessing
• Hand gesture detection, user identification
• Multi-model integration, real-time processing

• 🚀 Built and deployed scalable AI backend systems
• 📊 Designed high-performance APIs (async + caching + background jobs)
• ⚡ Reduced system latency by 40%
• 🔧 Integrated ML & LLM models into production pipelines
• 🛠️ Implemented monitoring, logging, and reliability improvements

Python • LangChain • OpenAI • Cohere • Google GenAI Embeddings • RAG • Semantic Search

FastAPI • Node.js • Express • REST APIs • Asyncio

Qdrant • pgvector • PostgreSQL • MongoDB

Docker • Redis • Celery • RabbitMQ • Nginx CI/CD • Prometheus • Grafana

React • Next.js • TypeScript

B.Sc. in Artificial Intelligence

Kafr-Elsheikh University (2019 – 2023)

GPA: 3.7

• 🔬 Building scalable RAG systems for real-world applications
• ⚡ Optimizing LLM latency, cost, and performance
• 🧠 Exploring advanced LLM architectures & agents
• 🤝 Open to collaboration on Generative AI & backend AI systems

📍 Alexandria, Egypt | 📞 +201206017652

• 🎯 Started as a self-taught developer
• 🧠 Passionate about making AI accessible and practical
• 🚀 Deployed multiple AI models from research to production
• 📚 Continuous learner - something new every day
• 🌍 Fluent in Arabic (Native) and English