At Schneider Electric, I engineered an equipment donation matching system that cut manual cross-referencing from hours to minutes. It combines web search, product catalog extraction, fuzzy matching, and NLP normalization to match equipment against a multi-country inventory.
I also built an impact reporting pipeline that consolidates project data into an Excel model and Power BI dashboard for Triple Bottom Line reporting. I created automated scraping pipelines for NGO directories in Egypt and Algeria.
At Cegedim, I contributed to Chameleon, a data masking product, assisted with CI/CD pipelines, and automated server configuration using Ansible.
In my projects, I built a RAG system that reduced hallucination rates from 70% to 12.7%, and an EGX market data pipeline using PostgreSQL and dbt. I also developed and evaluated deep learning models for stock price forecasting.
