Open to opportunities

Taj Eddine Marmoul

@tajeddinemarmoul

Message

Agentic AI engineer building MCP-based, auditable RAG and multi-agent systems.

France

Message

What I'm looking for

I’m looking for a role where I can ship MCP-based agent systems in production—auditable tool execution, validated actions, HITL approvals, and RAG grounding—while partnering with teams to harden deployment (security, logging, and reliability) for real users.

I’m an Agentic AI engineer focused on MCP-based production systems where tool execution is auditable, retrieval is grounded, and users can approve actions via human-in-the-loop (HITL). I build multi-agent workflows that favor bounded, validated actions over uncontrolled code generation.

At CNRS LAAS Toulouse, I built and deployed an MCP-based agentic web platform that turns natural-language requests into working web interfaces—supporting ~200 daily active users and generating 8,000+ forms and 3,000+ apps. I designed a multi-agent system with 4 MCP servers and 60 tools using FastMCP, FastAPI, React, and MongoDB, then hardened deployment with LDAP, TLS, rate limiting, input validation, audit logging, and HITL approval.

Previously, I delivered production-grade data and web systems: at Royal Air Maroc, I built a Dash app with PostgreSQL and Redis, reducing dashboard load time by 60%, and led a team of 5 to deploy versions for 100+ concurrent users using Docker, Nginx, and Gunicorn. Earlier, I built data-driven Unity educational games (analytics improved engagement by 40%) and led an AI security project using YOLOv5s and MobileFaceNet to reach 95% face-recognition accuracy, shipping a WebSocket full-stack app with role-based authentication for 50+ concurrent users.

Experience

Work history, roles, and key accomplishments

AI Platform Engineer Intern

CNRS LAAS Toulouse

Feb 2026 - Aug 2026 (6 months)

Built and deployed an MCP-based agentic web platform converting natural-language requests into working interfaces, supporting ~200 daily active users and generating 8,000+ forms and 3,000+ apps. Engineered a multi-agent system with 4 MCP servers and 60 tools and hardened production deployment with LDAP, TLS, rate limiting, validation, audit logging, and human-in-the-loop approval.

MCP fastAPI React Multi Agent Systems Human In The Loop MongoDB LDAP TLS

BI, Data & DevOps Intern

Royal Air Maroc

Jul 2025 - Aug 2025 (1 month)

Built a production Dash application for flight-operations data using PostgreSQL and Redis, reducing dashboard load time by 60%. Led a team of 5 to deliver web/local versions and automated ETL/KPI reporting, operating a Docker + Nginx + Gunicorn stack for 100+ concurrent users.

Dash PostgreSQL Redis ETL KPIs Docker Nginx Gunicorn Team Leadership Data Visualization

Unity Game Developer

Studyo.io

Oct 2024 - Jul 2025 (9 months)

Developed 7 educational Unity games with integrated analytics, improving student engagement metrics by 40% through data-driven iteration. Created optimized 3D assets and custom HLSL shaders, reducing game load times by 25% while preserving visual quality.

Unity C++3D Asset Optimization Shaders Performance Optimization Educational Gaming Rendering

AI & Full-Stack Intern

1337 Coding School

Jul 2024 - Present (1 year 11 months)

Led a team of 3 to build a real-time security system using YOLOv5s and MobileFaceNet, achieving 95% face-recognition accuracy across a multi-camera setup. Trained PyTorch models across 126 GPU-hours and delivered a WebSocket full-stack app with role-based authentication for 50+ concurrent users.

YOLOv5s PyTorch Computer Vision Multi Camera Systems WebSockets Full Stack Mobile FaceNet YOLO V5