• M.Eng. in Applied AI at DIT, Germany
• Bridging deep learning research with scalable software
• Building AI that is tested, evaluated and production ready

• Multi-agent workflows and tool calling with LangGraph
• Production RAG, Graph RAG and hybrid search with pgvector
• Grounding, structured output and LLM evaluation

• Multi-view learning with attention-based fusion
• CNN backbones and transfer learning
• Real-time detection and OpenCV pipelines

• Async APIs and WebSocket streams with FastAPI
• PostgreSQL design, Redis queues and caching
• Docker, Git and Linux workflows

Skills for AI coding agents to build products that feel truly Egyptian.

CopilotKit's open-source personal agent with a browser, terminal and files.

CopilotKit's open-source AI coworkers, each with a computer of its own.

Autonomous coding agent as an IDE extension, CLI and SDK.

The fast, Pythonic way to build MCP servers and clients.

Developed a deep learning multi-view classification approach for 38 mushroom species using three synchronized image views per sample. Designed a shared ResNet-18 visual encoder paired with learned attention-based feature fusion to dynamically weight information across views.

• Architecture: Shared ResNet-18 backbone + BatchNorm projection + learned attention pooling layer.
• Pipeline: End-to-end data preprocessing, augmentation, early stopping, and checkpointing.
• Evaluation: Comprehensive error analysis with learning curves, prediction plots, and confusion matrix heatmaps.

An intelligent resume analysis and optimization system utilizing autonomous agent workflows for targeted rubric scoring, ATS alignment, and version-controlled change recommendations.

• Agentic Loops: LangGraph multi-agent critique cycle with structured JSON validation.
• Semantic Matching: Vector embedding similarity and keyword gap analysis.
• Backend: Asynchronous FastAPI endpoints with PostgreSQL change tracking.

A scalable, asynchronous email orchestration and workflow automation platform engineered for high-volume transactional delivery, scheduling, and reliable queue processing.

• Async Architecture: Redis message brokering with dedicated task workers.
• Reliability: Rate limiting, exponential retry backoff, connection pooling, and dead-letter queues.
• Monitoring: Real-time delivery telemetry, latency tracking, and analytics.

Enterprise AI platform for cultural contracts, intellectual property governance, and grounded domain-knowledge retrieval with specialized Arabic LLM workflows.

• Knowledge Engine: Production RAG with pgvector hybrid search and citation verification.
• Graph RAG: Entity-relationship extraction for complex multi-party contract clauses.
• Safety: Prompt guardrails and hallucination evaluation frameworks.

An AI-assisted international education portal providing multi-criteria semantic matching between student academic profiles and global university degree requirements, streamlining admissions guidance.