Fahd Mekki

Designed and implemented a complete OCR-based invoice processing pipeline, including data extraction, analysis with large language models, and structured data generation. ◦ Fine-tuned LLaMA 3 8B via LoRA (F1 = 0.86) to improve accuracy and precision in invoice analysis. ◦ Integrated the optimized pipeline into the Odoo ERP system to automate invoice processing and streamline operations.

Designed an intelligent ERP system for sales forecasting and workforce planning, successfully used by store managers to optimize product and staff management. ◦ Developed and integrated XGBoost models (F1 = 0.91), trained on a large dataset from stores with various inputs such as date, season, and other factors, to accurately predict sales and workforce needs.

Developed a React Native app for vendors to manage orders and product data, replacing manual processes. Collaborated using Agile Scrum, integrated Magento 2 APIs, and enhanced vendor experience through real-time updates and intuitive UI.