Freelancer profile translated to English.

Description

I deliver AI systems in production, not POCs that gather dust.

I'm **Nathan**, an AI and agentic systems specialist. I help SMEs and mid-sized companies put AI to work: AI agents, RAG, automation, and copilots, integrated into your existing processes and tools, connected to your data, and reliable in real-world conditions. No need to overhaul everything: I graft AI onto existing systems where it creates value, in your CRM, ERP, email client, or business tools.

Specifically, I focus on:

LLM & RAG: your documents become a searchable knowledge base, with sourced answers (Claude, OpenAI, OS, pgvector)
Agents & orchestration: agents that execute real tasks in your tools, end-to-end (MCP, LangGraph, tool use)
Automation & workflows: your repetitive tasks and existing processes handled by reliable workflows connected to your software (n8n, Make, Python)
Cloud or on-prem: deployed where your constraints require, without your data leaving (Docker, AWS, GCP, Azure)
Audit & optimization: I identify where AI creates value and reduce costs by 30-70%

What sets me apart: a single point of contact from audit to deployment (the one who designs, codes, and deploys), anobsession with ROI(I'll advise against cases that aren't worthwhile), and measured systems (evaluations, safeguards, observability), not black boxes.

First chat is free: you'll leave with at least a frank opinion.

Languages

French
Native or bilingual
English
Conversational
Portuguese
Basic
Spanish
Basic

Workplace preferences

Can work on-site

Paris (up to 50km)

Kyndryl
AI Architect Intern
DIGITAL AND IT
June 2026 - Today (1 month)
Brno, Czechia
• • Built a full-stack enterprise solution for intelligent storage-capacity management and anomaly detection across heterogeneous multi-OS fleets; orchestrated LLM agents with LangGraph to explain, simplify, and prioritize detected anomalies.
• • Designed a three-engine anomaly-detection system (statistical Tukey/Theil-Sen, ML Histogram Gradient Boosting + Isolation Forest, hybrid); Python 3.12 / FastAPI backend and Next.js 16 / React 19 client (TypeScript) with interactive dashboards and automated PDF reports.
Langchain Python LangGraph Typescript FastAPI
Stellantis
Engineering Apprentice – Observability & Generative AI
DIGITAL AND IT
September 2024 - Today (1 year and 10 months)
Sochaux, France
• • Drove the shift from infrastructure observability to agentic Generative AI: designed a Model Context Protocol (MCP, fastmcp) infrastructure decoupling LLMs from information-system tool execution, with enterprise middleware (secure secrets, LLM-call observability) and a Zero Trust / SSO MCP proxy in Node.js.
• • Designed an end-to-end RAG chain and a Python service for idempotent synchronization between technical GitHub repositories and the Kibana AI Assistant knowledge base.
• • Engineered a native Ruby Logstash filter that calls an LLM during log ingestion (anomalies as structured JSON); CMSDB extraction pipelines and dynamic Ansible-inventory generation via the Elasticsearch REST API.
AI Agent RAG AI Automation Python IT-Security