Mahmoud Saleh

Production-style multimodal AI system for image and document analysis with grounded explanations using vision and language models.

AI Engineer / Cloud-AI Developer portfolio featuring ML, Azure, FastAPI, and RAG projects.

AI Engineer & Cloud-AI Developer | Building ML systems, FastAPI microservices, Azure deployments, and RAG applications.

Backend-first job market intelligence platform that ingests job postings, analyzes skill demand and competition, and produces explainable daily recommendations and weekly market insights.

An always-on, cloud-native system that ingests live public incident feeds, performs real-time ML classification and LLM-based summarization, and exposes actionable risk signals with full observability under bursty traffic.

End-to-end MLOps system for churn prediction with MLflow model registry, alias-based deployment, shadow evaluation, and production-grade inference.

No description provided.

Serverless invoice extraction API using Azure Document Intelligence and Azure Functions. Upload a PDF invoice and receive normalized JSON output including line items, totals, dates, and vendor details.

Production-grade customer segmentation pipeline built on Azure (Blob Storage, Data Factory, Azure ML, Batch Endpoint). Includes end-to-end data engineering, feature engineering, K-Means model training, and scalable batch inference.

Production-ready Retrieval-Augmented Generation (RAG) microservice using FastAPI, ChromaDB, SentenceTransformers, Grok (xAI), and Docker. Supports TXT/PDF ingestion, vector search, and LLM-based question answering.

A clean, well-structured PyTorch project where I trained a ResNet18 model on the Intel Natural Scenes dataset. The repo includes modular code, a simple training pipeline, full documentation, and a reproducible process that reflects how I approach real ML work.

FastAPI + Docker inference service for my fine-tuned ResNet18 Intel Natural Scenes classifier. Upload an image and get real-time scene predictions with confidence scores.

This notebook demonstrates the core building blocks of deep learning using PyTorch:

Retail analytics case study using Python, SQL, and Tableau. Analyzes sales and profit trends, identifies how discounting drives margin volatility, and simulates what-if scenarios showing how small discount reductions could increase total profit by 10–25%.

Data analysis of Bellabeat smart device data to uncover activity, intensity, and sleep trends — Google Data Analytics Capstone Project.

Implement an image classification model using transfer learning with pretrained architectures (VGG16 or ResNet18).

Forecast use of a city bikeshare system