Taha KHOMSI

Data & AI Automation Engineer with experience designing data platforms, enterprise workflow automation, and data governance solutions across finance, energy, and logistics. I build scalable data warehouses, automate business-critical processes, and integrate data systems into operational workflows — combining data engineering, BPM (Appian), and applied AI to deliver high-impact automation. 🏦 UFF (Union Financière de France – Banking & Insurance) — 2+ years (2023–Present, Paris) Led the full retirement of Salesforce across all UFF agencies in France, migrating 12 reports into Appian and replacing Salesforce Chatter with an in-app discussion channel (CRM moved to a separate platform). Built the replacement Appian reporting solution: automated regulatory and operational report generation with a self-service interface for filtering and CSV export, removing manual report production. Designed and maintained a centralized enterprise data warehouse consolidating 5+ business data sources into a unified analytical model aligned with enterprise governance standards. Optimized database performance through query-plan analysis (EXPLAIN), targeted indexing, and a trigger-based precomputation strategy — materializing expensive computed values into a dedicated table at write-time to offload read-time query load. Implemented GDPR-compliant data lifecycle and document-retention automation in production. Developed workflow-driven automation systems in Appian supporting cross-team business processes. 📦 Touton (Logistics & Supply Chain) — 2 months (2023, Bordeaux – Gutilab) Supported development and maintenance of a warehouse tracking system for coffee supply-chain traceability. Contributed to workflow maintenance, system testing, and data validation in a production logistics environment. ⚡ Romande Énergie (Energy & Meteorology) — 4–5 months (2023, Remote) Designed an automated monitoring system combining meteorological data and electricity-price signals. Built pipelines integrating multiple external APIs with differing refresh frequencies, triggering real-time alerts for energy infrastructure operations and enabling proactive decision-making.

📡 Bouygues Telecom (Telecommunications) Built an automated conformity-control system using YOLOv5 to detect and classify 6 types of telecom network equipment, reaching 97% mAP on the test set — trained on a 7,000-image dataset I collected, annotated, and cleaned — then wrapped the model in a Python application that scores conformity automatically, cutting manual effort on repetitive checks by ~95% and processing time by ~90%. Designed an end-to-end pipeline combining object detection, OCR, and data processing (YOLOv5, PaddleOCR, OpenCV, regex) to verify technical PDF reports against client requirements, optimizing and validating each sub-module with Python scripts. Automated the historization and reporting of alarms and alerts (Python, MySQL, Power BI) on top of the Data Analyst team's dashboards, and built visualizations tracking each model's performance to support team-lead decisions. Reviewed the state of the art and benchmarked candidate techniques to select the best-fit approach.

Built an ETL pipeline consolidating multiple company data sources (websites, databases, pay slips) and derived 5 KPIs to measure employee productivity. Ran exploratory data analysis to find salary-correlated features, then built a linear-regression model predicting employee salary from the 5 KPIs — retrained continuously on new pay slips — and surfaced results in a Power BI dashboard. Developed a responsive, manager-only module (Java Spring Boot, Angular) in the company website showing employee productivity and predicted monthly salary.

Built a full-stack HR leave-management web application (Angular, Spring Boot) end to end — requirements, UI design, implementation, testing — alongside the consulting team.

Introductory internship at Morocco's central bank: exposure to ERP systems, procedural standards, and core concepts of solvency and risk management