I'm an AI/LLM Engineer and Python Backend Developer based in Abu Dhabi, UAE, with 3+ years of experience building production-grade APIs, intelligent chatbot systems, and cloud-native microservices.

My core strength is bridging AI capabilities with scalable backend architecture — from designing multi-agent platforms and RAG pipelines to optimising database schemas serving millions of records. I delivered a 40% improvement in CRM sales efficiency and built real-time property data sync across major UAE listing portals at Aqary International Group.

• 🤖 AI/LLM: LangChain agents, RAG pipelines, OpenAI GPT-4, Hugging Face Transformers
• 🏗️ Backend: Production-grade Django, Flask, FastAPI microservices on Azure & AWS
• 🔬 Research: ECG biometric authentication at CardioAuth
• 🌱 Learning: Advanced Python backend engineering, MLOps, cloud-native infrastructure
• 📍 Location: Abu Dhabi, UAE · +971 58 560 9293

Jan 2024 – Present · Abu Dhabi, UAE

Aqary is a UAE-based real estate technology company. As their Backend & AI Engineer, I own the AI layer and core backend infrastructure powering their CRM, property portals, and internal tooling.

🤖 AI & LLM Work

• Built production AI chatbots using LangChain agents and chains backed by OpenAI GPT-4 and Google Gemini; implemented tool-calling, memory management, and context windowing for complex multi-turn customer support conversations
• Designed RAG (Retrieval-Augmented Generation) pipelines using LangChain + pgvector: chunked and embedded property documents, stored vectors in PostgreSQL, and retrieved semantically relevant context for LLM responses — reducing hallucinations and improving answer accuracy
• Integrated Hugging Face Transformers for sentiment analysis on customer feedback, multi-label text classification for property listings, and automated Arabic/English content generation using pre-trained and fine-tuned models
• Leveraged OpenAI Embeddings API to power semantic search across property listings, replacing keyword search and improving search relevance

⚙️ Backend & Infrastructure

• Engineered RESTful APIs in Flask and Django following OpenAPI spec; applied rate limiting, request validation with Pydantic, and versioning — achieving 99.9% uptime across all endpoints
• Implemented JWT-based authentication with RBAC, refresh token rotation, and OAuth 2.0 for third-party integrations
• Led Docker + Kubernetes deployment pipelines on Azure with HPA, zero-downtime rolling updates, and environment parity across dev/staging/production

🔗 CRM & Portal Integrations

• Architected Zoho CRM integrations using REST APIs and webhook automation, eliminating manual data entry for 20+ sales agents and boosting sales team efficiency by 40%
• Built XML-feed and REST integrations with Bayut and Property Finder portals, enabling real-time bi-directional property data sync across thousands of active listings

📊 Database Optimisation

• Optimised PostgreSQL and MySQL schemas for millions of records: redesigned indexes, rewrote N+1 query patterns using select_related/prefetch_related, and added Redis query caching — cutting average API response time by ~35%

Dec 2024 – Present · Remote — Abu Dhabi, UAE

• Engineered a full-featured e-banking backend (Flask + PostgreSQL) with RBAC, JWT auth, account-to-account transfers, real-time transaction logging, and a fraud-monitoring admin dashboard handling 1,000+ daily transactions
• Implemented async task queues with Celery + Redis for background transaction processing, notification dispatch, and scheduled financial report generation — keeping API response times under 200ms
• Built a purchase-order management system with multi-level approval workflows, supplier tracking, and payment-status tracking via a clean REST API integrated with a React frontend
• Containerised all services with Docker Compose; wrote CI/CD pipelines with GitHub Actions (lint → test → build → deploy); maintained full test coverage with pytest

Team Project · Working MVP

An AI-powered system that converts plain-text prompts into production-ready database schemas, auto-generates SQL migrations, and populates tables with realistic dummy data — zero manual fixture writing.

What it does

• Accepts a plain-text description like "a multi-tenant SaaS platform with users, subscriptions, and invoices" and outputs a fully normalised, migration-ready schema with correct data types, constraints, foreign keys, and indexes
• Automatically generates realistic dummy data tailored to each column type (names, emails, prices, dates, enums) using parameterised INSERT statements

Technical Architecture

• Designed the core LLM pipeline using OpenAI GPT-4 via OpenRouter — the system is fully model-agnostic: users can select GPT-4, Claude, Mistral, or any supported model at runtime
• Engineered structured prompt chains that parse free-text descriptions into normalised ER models, then emit valid CREATE TABLE SQL for PostgreSQL and MySQL
• Architected a microservices backend with FastAPI: separate services for prompt parsing, schema generation, migration execution, and data seeding — each independently deployable
• Offloaded long-running generation jobs to Celery + Redis task queues, keeping API response times under 300ms while processing complex multi-table schemas asynchronously
• Containerised all microservices with Docker Compose: inter-service networking, volume mounts for DB persistence, environment-based config for multi-DB target support

Stack: FastAPI OpenAI GPT-4 OpenRouter Celery Redis PostgreSQL MySQL Docker Compose

Production-Ready SaaS · Python & TypeScript SDKs + CLI

A comprehensive Flask-based platform enabling companies to create, train, and deploy intelligent chatbot agents with advanced NLP capabilities and enterprise-grade scalability — plus a full SDK and CLI ecosystem for developer integration.

Platform Highlights

• Multi-Agent Management: Create specialised chatbots for different business functions (support, sales, onboarding)
• Custom Training: Upload documents (PDF, DOCX, CSV, XML, JSON), connect databases, integrate websites
• Enterprise Scale: Multi-tenant architecture with complete data isolation per client
• Advanced Analytics: Lead conversion tracking and agent performance monitoring
• AI Integration: Gemini API, custom embeddings, OCR processing
• Security: Multi-level authentication, data encryption, audit logging

Knowrithm SDKs & CLI

• Designed type-safe Python and TypeScript SDKs with JWT/API-key auth, automatic retries, streaming response support, and multipart file uploads — reducing client integration time by ~60%
• Built a full-featured CLI with interactive wizards, agent lifecycle management, document ingestion, analytics querying, fuzzy search, and multi-format output (JSON, table, CSV, YAML)
• Integrated AWS S3 for document storage and AWS EC2 for compute; added MinIO support for on-premise and hybrid cloud deployments
• Implemented async polling for long-running document-processing jobs, robust pagination, and typed exception handling

Backend Architecture

• Backend: Flask + SQLAlchemy ORM + connection pooling
• Queue System: Celery workers for asynchronous document processing
• Cache Layer: Redis for high-performance caching and message brokering
• Database: PostgreSQL with multi-tenant data isolation

Stack: Flask PostgreSQL Redis Celery JWT Vector Search Gemini API AWS S3 AWS EC2 MinIO Python SDK TypeScript SDK

Freelance Project · Production Grade

A comprehensive digital banking backend engineered to handle secure financial operations at scale.

• Full RBAC + JWT authentication with refresh token rotation and multi-factor auth readiness
• Account-to-account transfers with real-time balance updates, configurable transaction limits, and escrow support
• Real-time transaction logging with audit trail for every financial operation
• Fraud-monitoring admin dashboard handling 1,000+ daily transactions with anomaly alerts
• Async task processing via Celery + Redis for notifications, report generation, and background reconciliation — API response times under 200ms
• Full CI/CD pipeline with GitHub Actions (lint → test → build → deploy) and pytest test coverage

Stack: Flask PostgreSQL Redis Celery JWT Docker Compose GitHub Actions pytest

Research & Development

An advanced biometric authentication system using electrocardiogram (ECG) signals and deep learning for secure identity verification.

• Real-time ECG signal analysis with neural network classification
• End-to-end ML pipeline: preprocessing → training → inference
• RESTful API built in Flask for integration with external systems
• Research focus on biometric authentication in distributed systems

Stack: Python TensorFlow PyTorch Signal Processing Flask

B.Sc. Computer Science — Summa Cum Laude (GPA 3.5 / 4.0) Assiut University · Graduated Jul 2023 · Assiut, Egypt

• Competitive programming: Codeforces contestant; participated in ECPC (Egyptian Collegiate Programming Contest)