T--.Multi-disciplinary work across code, design & AI
Challenge:
Academic project for ESPRIT (5DS8): make urban traffic adaptive with Eclipse SUMO and Multi-Agent Reinforcement Learning — each intersection is an autonomous agent learning to cut congestion, protect pedestrians and prioritize emergency vehicles across 2x2/3x3 grids, Monaco and Tunis.
Services:
Role:
Simulation & Environment Lead: SUMO network scenarios, TraCI metrics pipeline, fixed-time / Max-Pressure baselines, and the evaluation harness behind the live control-room dashboard.
Status:
Done: SUMO networks (grids, Monaco, Tunis), TraCI metrics, baselines and the live dashboard. In progress: MARL benchmark vs fixed-time and Max-Pressure.



Challenge:
Academic project for Yale University: automate the Fellowship Board grant panel with n8n — document intake and parsing, LLM-assisted triage and scoring, reviewer assignment, and status notifications, keeping human-in-the-loop decisions.
Services:
Role:
Designing n8n pipelines for application intake, doc parsing, LLM screening and ranking, and panel review workflow.
Status:
Delivered: document intake + parsing, multi-agent evaluation pipeline, LLM scoring calibration, and the reviewer loop — end to end in n8n.


Challenge:
Give insurers a precise, auditable climate-risk score for any French address in minutes: crossing official data (Géorisques, BDNB, IGN) with generative AI to measure risk before the disaster strikes, across flood, seismic, radon and land-movement hazards.
Services:
Role:
AI Engineer: built the platform end to end: Mapbox 3D map engine with building extrusions, multi-hazard risk scoring from Géorisques data, generative-AI recommendations, flood simulation timeline, batch portfolio diagnostics, and automated PDF report generation.




Challenge:
Build ALIA for Vital Laboratoire: a conversational AI avatar that greets visitors like a human front desk — answering in their language (French, Darja, English, Spanish, Italian), guiding them to the right service, and handing off to staff when needed.
Services:
Role:
End to end: avatar conversation design, multi-language flows, visitor profile updates, delegate training dashboards (sessions, scores, step-by-step performance), and a ticketing panel for support escalation.



Challenge:
Build a SaaS platform that analyzes and visualizes billing data: turning complex financial information into actionable insights through an intuitive CRM interface.
Services:
Role:
Led product design and full-stack development of the billing analytics CRM, focusing on data clarity, user workflows, and a seamless SaaS experience.


Challenge:
Design a CRM tailored for real estate professionals: managing properties, client relationships, and sales pipelines in one cohesive platform.
Services:
Role:
End-to-end design and development of the real estate CRM, from user research and information architecture to final interface and pipeline workflows.
Challenge:
Build an AI-powered travel CRM platform during an AI engineering internship: integrating intelligent trip planning, itinerary management, and client engagement features for a seamless travel experience.
Services:
Role:
AI Engineer Intern: contributed to the development of an intelligent travel CRM, working on AI features, data pipelines, and user-facing CRM interfaces for ctravel.


Challenge:
Design and develop an academic project at ESPRIT that stands out: achieving 2nd place at Bal de Projet 2025 with a compelling concept, polished execution, and strong presentation.
Services:
Role:
Designer & Developer: contributed to the end-to-end design and development of Matchupz, an award-winning academic project recognized at ESPRIT's Bal de Projet 2025 competition.




If you’re still here is for a good reason
I help make your projects clear and hard to forget.