AI Engineer & Cloud-AI Developer

I design and build end-to-end AI systems: training models in PyTorch, deploying them on Azure, and serving them through clean, production-style APIs.

View Projects Contact Me
Cairo, Egypt · Open to Remote Python · Azure · ML · FastAPI · Docker

What I do

  • ✅ Train and fine-tune deep learning models in PyTorch
  • ✅ Serve ML models via FastAPI microservices
  • ✅ Deploy pipelines on Azure ML & Azure Functions
  • ✅ Build RAG systems with vector databases & LLMs
GitHub LinkedIn

Core Skills

Applied machine learning, cloud deployment, and production-ready Python services.

AI & Machine Learning

  • PyTorch, torchvision
  • CNNs, transfer learning, classification
  • Classical ML (K-Means, feature engineering)
  • Model evaluation, training workflows

Cloud & MLOps

  • Azure ML (Batch Endpoints, model registry)
  • Azure Functions (serverless APIs)
  • Azure Document Intelligence (invoice extraction)
  • Infrastructure-as-code (YAML for jobs & endpoints)

APIs & Backend

  • FastAPI for ML and RAG microservices
  • REST API design, health endpoints
  • Docker & Docker Compose for deployment
  • Environment management, configuration, logging

Data & Engineering Background

  • Python, SQL, Pandas, ETL workflows
  • 10+ years in embedded & automotive software
  • System design & requirements analysis
  • Clean code, documentation, reproducible repos

Featured Projects

Selected projects that show how I combine ML modeling, Azure deployment, and production-grade FastAPI services.

Azure Customer Segmentation Pipeline

K-Means · Azure ML Batch Endpoint · Azure Data Factory

Built an end-to-end customer clustering engine on Azure that segments online retail customers using K-Means and a production-grade batch inference pipeline.

  • Feature engineering for recency, frequency, monetary value, and more
  • Model training & registration in Azure ML
  • YAML-based Batch Endpoint & deployment definitions for reproducible infra
Cloud ML Pipeline
View on GitHub

Azure Serverless Invoice Extraction API

Azure Functions · Document Intelligence · CI/CD

Developed a serverless API that accepts PDF invoices and returns normalized JSON with header fields, totals, vendors, and line items using Azure Document Intelligence.

  • HTTP-triggered Azure Function with a dedicated extraction service layer
  • Normalization of Azure DI output into clean, consistent JSON
  • GitHub Actions pipeline with tests and post-deploy health checks
Document AI
View on GitHub

RAG Microservice with FastAPI & ChromaDB

FastAPI · SentenceTransformers · ChromaDB · Grok

Implemented a Retrieval-Augmented Generation microservice that ingests TXT/PDF documents, indexes them into ChromaDB, and answers questions using embeddings and the Grok LLM via a FastAPI `/ask` endpoint.

  • Offline ingestion pipeline for chunking, embedding, and vector storage
  • Online query pipeline with retrieval + grounded answer generation
  • Dockerized service with environment-based LLM configuration
LLM / RAG
View on GitHub

Intel Natural Scenes API

FastAPI · PyTorch · Docker

Built a FastAPI microservice that exposes a fine-tuned ResNet18 model for 6-class scene classification, returning predictions and confidence scores for uploaded images.

  • `/predict` endpoint for inference and `/health` for monitoring
  • Clean separation of API, model loading, and preprocessing utilities
  • Fully containerized with Docker for cloud deployment
Inference API
View on GitHub

Intel Natural Scenes Classifier

PyTorch · ResNet18 · Transfer Learning

Trained a ResNet18-based classifier on the Intel Natural Scenes dataset, using a two-phase transfer learning strategy to adapt ImageNet features to 6 outdoor scene categories.

  • Phase 1: train only the classification head with frozen backbone
  • Phase 2: fine-tune the last ResNet block with differential learning rates
  • Exported a compact checkpoint used directly by the API project
Model Training
View on GitHub

Flower Image Classifier

PyTorch · VGG/ResNet · CLI Tools

Implemented a transfer-learning-based classifier for flower species, with scripts for training, saving checkpoints, and running predictions from the command line or a demo notebook.

  • Configurable backbone (VGG16 or ResNet18) and classifier head
  • Training & prediction scripts with GPU/CPU support
  • Top-K prediction visualization in a polished demo notebook
Deep Learning
View on GitHub

Experience

Strong engineering background with leadership roles in embedded systems, now combined with applied AI and cloud development.

AI Engineer / Cloud-AI Developer

2024 – 2025

  • Designed and deployed end-to-end AI systems: ML models, FastAPI microservices, Azure Functions, and batch pipelines.
  • Developed LLM/RAG microservices using FastAPI, ChromaDB, SentenceTransformers, and Groq.
  • Built Azure ML customer segmentation pipeline with feature engineering, model registry, and batch endpoints.
  • Created serverless invoice extraction API using Azure Document Intelligence with CI/CD, testing, and monitoring.
  • Developed PyTorch deep learning models and Dockerized inference services for production deployment.

Software Technical Leader – Fuel Cell Compressors

Garrett Advancing Motion · Prague · Jun 2024 – Feb 2025

  • Technical leader for the inverter software team developing embedded software for Fuel Cell Compressors.
  • Owned development, integration, testing, and release of embedded software according to ASPICE and ISO 26262.
  • Led software architecture, requirements, and feature planning, working closely with global OEM customers.
  • Managed issue analysis, debugging, and resolution; coordinated cross-functional technical discussions.
  • Supported RFI/RFQ phases by estimating workload, defining scope, and identifying technical risks.

Senior Principal / Senior MBD Engineer – Electrified Mobility & Wiper Systems

Valeo · Cairo · 2014 – May 2024

  • Technical leader for Electrified Mobility software (i-StARS, 48V GMG): model-based design, code generation, testing.
  • Designed & validated embedded software using Simulink, Embedded Coder, and MIL/SIL workflows.
  • Certified MBD reviewer: ensured process compliance, performed quality reviews, and assessed technical risks.
  • Led software development cycles for Adaptive Wiper Systems: planning, requirement analysis, issue tracking, and audits.
  • Implemented safety requirements and testing for ISO 26262 ASIL A–C compliance across multiple projects.
  • Developed automation tools (MATLAB/Simulink) that reduced development effort and improved debugging efficiency.

Certifications & Learning

Structured learning paths that support my shift into AI & cloud engineering.

Formal Programs

  • Google Data Analytics Professional Certificate
  • Udacity AI Programming with Python Nanodegree
  • AWS Cloud Fundamentals

Self-Directed Learning

  • Azure ML, Azure Functions & Azure Document Intelligence
  • FastAPI, Docker, vector databases, and LLM tooling

Contact

Open to remote roles, freelance work, and collaborations on AI/ML projects.

Get in touch

If you're looking for someone to design, build, and deploy ML systems and AI-powered APIs, I'd be happy to connect.

Links