Youssef Cherif

Build and maintain enterprise applications for manufacturing costing, quality and action management used across the AVOCarbon group; work directly with the Group CEO and senior stakeholders to turn operational requirements into working, auditable software. • Design FastAPI services, PostgreSQL data models, REST APIs and React interfaces; manage schema migrations, scheduled jobs, email workflows, Docker environments and Azure deployments. • Delivered an action-management platform with automated reminders, priority scoring, escalation rules and hierarchy-based reporting, including duplicate-data cleanup across group entities. AI-Assisted Costing & RFQ Automation Platform (key project) • Built a drawing-first RFQ workflow that converts customer PDF drawings and commercial inputs into structured bills of materials, supplier component costs, MOST assembly estimates and preliminary manufacturing costs, using specialised AI agents integrated through an MCP layer with asynchronous callbacks and correlation IDs. • Developed a traceable, Decimal-precision cost engine covering materials, logistics, direct labour, variable overhead, fixed overhead and fees, with firm and preliminary costing modes and explicit exclusion of unresolved components instead of silent zeros. • Implemented a deterministic financial-planning module: Y-1 to Y6 projections, sales and EBITDA, trade working capital, CAPEX and tooling schedules, depreciation, cash flow, NPV and an automatic Y0 selling-price solver. • Reached 198 passing automated backend, API, financial and regression tests; validated the complete workflow on a real production RFQ and cross-checked AI-generated costs against historical costing files. Supply Chain Intelligence Platform — MLOps & AI (final-year engineering project, 2026) • Engineered a LightGBM demand-forecasting model for 52 items over a 12-week horizon, reducing WAPE from 81% (LSTM baseline) to 47%, combined with confirmed-order logic covering 29 items and 16.7 million units. • Implemented PSI drift detection, four automated governance rules, champion/challenger selection, an MLflow model registry and operational alerts, with weekly automated runs via APScheduler. • Built React/Recharts dashboards for forecasts, WAPE, PSI and governance events, and a Qwen2.5 LLM assistant using ReAct tool calling across six live database tools for stock, forecast and alert queries.

Développement d’un script Python automatisé pour surveiller un serveur et détecter l’arrivée de nouveaux fichiers Excel (schedule). Fusion et nettoyage des fichiers Excel, préparation des données pour l’intégration. Stockage des données nettoyées dans une base PostgreSQL pour exploitation analytique. Conception et réalisation d’un dashboard interactif Power BI pour visualiser les résultats et faciliter la prise de décision. Optimisation des processus de collecte et d’analyse de données, améliorant la productivité et la fiabilité du reporting.

Conception et réalisation d’une application web (React.js, Flask) et mobile (React Native) permettant la classification et le suivi de la qualité de l’eau potable. Mise en place d’un backend en Python Flask et d’une base de données PostgreSQL pour la gestion et le stockage des données. Intégration d’algorithmes de traitement et de classification afin de faciliter la prise de décision et le contrôle de la qualité. Collaboration en équipe dans un cadre scientifique appliqué aux problématiques de santé publique.