Ali Mahran

I’m an ML/AI Engineer with +3 year experience, focused on building end-to-end machine learning systems and deploying AI that performs reliably in production. I enjoy turning messy data into measurable outcomes through data pipelines, feature engineering, and strong evaluation.

At Prop-AI (May 2024 - Present), I architected and maintained scalable data pipelines and ETL workflows to ensure dependable ingestion, transformation, and availability for downstream ML systems. I designed and deployed end-to-end ML workflows (training, evaluation, and production deployment), built cost-efficient multi-agent AI using smaller specialized models, and implemented monitoring, alerting, and drift detection for continuous improvement.

Before that, at DMS (July 2023 - May 2024), I worked with time series using statistical techniques and built an Arabic sentiment classifier using XGBoost and Random Forest. I also developed an Arabic conversational chatbot with an open-source LLM and implemented Retrieval Augmented Generation (RAG) using vector databases to improve retrieval and generation quality.

My graduation project, “Teeth Segmentation,” involved implementing and optimizing Attention U-Net for semantic segmentation on the TUFTS dataset, and I submitted a manuscript titled “Automatic Teeth Segmentation Using Attention U-Net” to Springer for journal publication. I’m driven by efficiency, scalability, and real-world applicability—especially with LLM-based and agentic architectures.