Freelancer profile translated to English.

Description

I design and industrialize auditable AI systems for SMEs, mid-market companies, and large corporations in France and Switzerland. Multi-agent systems, RAG, code-based automation, custom integrations.

My approach is based on three pillars: auditability by default, systematic measurement, and industrialization up to production. I don't deliver demos, I deliver systems that work.

Specifically, I can help you to:

Architect a multi-agent system (4 to 10 agents) with n8n, LangChain, or custom code orchestration
Implement an advanced RAG pipeline (CRAG, Self-RAG, GraphRAG, Agentic RAG) with guardrails and continuous evaluation
Integrate LLMs into production (OpenAI, Claude, Mistral, Gemini) with GDPR anonymization, monitoring, and auditing
Connect your business stack (Pennylane, Odoo, Notion, CRM, ERP) to agents that make traceable decisions
Refactor a failing backend (Python, TypeScript, Rust) or audit an existing architecture

Recent achievements:

Multi-agent OSINT platform: 8 agents, 90+ sources, investigation time reduced from 4 hours to less than 2 minutes.
SMAD, an auditable AI decision support system for finance: 4 agents in parallel, two-pass corrective RAG, dynamically weighted scoring, SHA-256 audit of each decision.
B2B AI SaaS platform for Marubeni: solo design and development, from architecture to VPS deployment. Five functional hubs, 11 business domains.

What sets me apart: my real job is to solve complex problems. I come with a methodology that covers the entire cycle, from scoping to production. Architecture design, code piloting, deployment, results measurement. You have a single point of contact to unblock situations.

Typical profiles: CEO or CFO of an industrial or financial mid-market company, CTO of a tech scale-up, innovation manager, SaaS founder.

First scoping call offered, 45 minutes, response within 24 hours.

Industry field of expertise

Languages

French
Native or bilingual
Turkish
Native or bilingual
English
Basic

Workplace preferences

Can work on-site

Annecy (up to 50km)

MINDSET CONNECT
AI & Automation Architect - Multi-Agent Systems
BANKING AND INSURANCE
January 2026 - April 2026 (3 months)
Paris, France
Design and development of the SMAD (Smart Multi-Agent Decision) system, an intelligent financial arbitrage platform.

Architecture orchestrating 4 specialized agents in parallel execution (LOGOS - logical analysis, ACTION - execution guidance, ETHOS - ethical compliance, HOLOS - holistic vision), whose results are merged, aggregated, and then validated by a two-level gate system.

The decision-making core relies on MindScore, a trust scoring algorithm that dynamically determines the processing path: direct pass, enrichment through two-pass RAG with intelligent query rewriting, web fallback, or human escalation (Human-in-the-Loop).

The RAG pipeline integrates a two-pass relevance verification mechanism with dedicated LangChain agents and structured parsers. If the first pass is insufficient, a rewriting agent reformulates the query before a second attempt, with fallback to web search and human escalation as a last resort.

Each decision is tracked via an audit trail persisted in PostgreSQL, ensuring regulatory compliance in the financial sector.
n8n RAG Automatisation API REST NLP
Marubeni International (Europe)
Full Stack Developer
SOFTWARE PUBLISHING
May 2025 - November 2025 (6 months)
75009 Paris, France
AI Hub for prompt creation, versioning, and collaborative planning
Solution that creates prompts with history, comparison, and restoration
Implementation of the Prompt Scheduler: scheduled sends (email, Slack, AI API)
Job offer generator and AI CV analysis
Hierarchical folder management with READ / EDIT permissions
Secure authentication, user management
Intégration d’API externe intelligence artificielle Prompt engineering JavaScript / TypeScript PostgreSQL
French Tech Mont-Blanc
Python Developer / Full Code Automation
SOFTWARE PUBLISHING
May 2025 - May 2025
Passy, France
🎯 finalist of the Project carried out as part of a hackathon organized by La French Tech

SEO writing automation through AI + Notion integration.

Creation of automation tools in Python (workflow without using platforms like: n8n, zapier, make)
Integration of the Mistral AI API and the Notion.so API
Secure API-first architecture with environment variable management
Automated generation of optimized content in French
Prompt Chaining
Skills:Python, API integration (Notion, Mistral), Generative AI, SEO, cost reduction, accelerated production, scalability
intelligence artificielle Python Intégration d’API externe SEO Automatisation