Pierre Rondeau

// status: open to work

AI ENGINEER

I put agentic systems into production. Not demos.

  • Nantes or remote EU
  • Available immediately
  • Full-time
  • FR / EN
28+
companies shipped for
3x
Anthropic certified
6+
MCP servers in production
5 yrs
writing production code

What I Build

Proof of work, not proof of talk.

AI-powered prospecting system : scraping, scoring, and personalized outreach via MCP agents.

Leads processed
100+
Pipeline
Automated
Claude Code MCP Unipile API n8n Notion
  • /Python
  • /TypeScript
  • /Anthropic API
  • /OpenAI API
  • /OpenRouter
  • /LiteLLM
  • /MCP
  • /RAG
  • /pgvector
  • /Qdrant
  • /Pinecone
  • /FastAPI
  • /Django
  • /Next.js
  • /React
  • /PostgreSQL
  • /Supabase
  • /Redis
  • /Docker
  • /AWS
  • /GitHub Actions
  • /GitLab CI
  • /n8n

What I'm looking for

The kind of role where I am useful from the first sprint.

01

Building agents that act

Orchestrating complex business workflows, tool use, guardrails and error recovery. Not a demo chatbot.

02

Industrialising RAG

Ingestion, hybrid search, reranking, continuous evaluation. Robust and measurable, not a notebook that worked once.

03

Building GenAI platforms

Unified APIs, multi-LLM, provider routing, failover and cost control. Exactly what I built for Coffee Leads.

04

Wiring models into the business

MCP, CRM, ERP, document stores. An assistant is only useful once it reaches the company’s real systems.

05

Understanding how it works

Not just consuming APIs: architecture, evaluation, model limits, FinOps. I want to know why it breaks.

06

A team where engineering is the point

Craft, code review, sharing and training. I already do it: 28 companies trained, ClaudeHub, a Gen AI hackathon co-organised.

Stack & Tools

What I use daily to build, automate, and ship.

LLM & Agents

Anthropic API Claude in production: agents, pipelines, business tools
OpenAI API Second provider in multi-model pipelines
OpenRouter Multi-provider routing and model failover
LiteLLM Abstraction layer and cost control
MCP Bespoke servers: Notion, CRM, LinkedIn, Calendar, Dynamics
Claude Code Subagents, skills and hooks orchestrating daily work
LangGraph Migration under way from my in-house orchestrator
DeepEval LLM output evaluation, being adopted

RAG & Data

pgvector Embeddings stored in Postgres, in production
Pinecone Dedicated vector store for semantic search
Qdrant Self-hosted vector store
Hybrid search Lexical and vector combined, then reranking

Langages

Python Agent engines, data pipelines, APIs
TypeScript Type safety across the web stack and MCP servers

Backend

FastAPI Async APIs exposing agents
Django Web apps and REST APIs, from degree work to client projects
PostgreSQL Main relational database, with vector extension
Supabase Auth, managed Postgres and realtime
MongoDB Document storage on startup projects
Redis Caching and job queues

Frontend

Next.js React applications and product interfaces
React Product interfaces and dashboards
Vue.js Business app front-ends in agile teams
Astro Content sites, static-first
Tailwind CSS Utility-first styling, fast and consistent

Infra & Ops

AWS Lambda and S3: applications deployed and operated in production
Docker Containers and Docker Compose, on every project
CI/CD GitHub Actions, GitLab CI and CircleCI
Observabilité Monitoring and structured logging on a managed platform
Vercel Edge deployment for web projects
Railway Hosting for services and MCP servers

Automation

n8n Business process automation, delivered as on-site training
Make No-code automation, and migrations to n8n
Pierre Rondeau

About

I didn't take the straight path. Dropped out of high school, bounced between cities, jobs and false starts. Worked in a mountain restaurant where the chef was an alcoholic and the manager lost his temper daily. Spent months alone in a room wondering what I was supposed to do with my life.

Then someone asked: "Have you tried coding?" That one question changed everything. In four months I went from zero to accepted into a dev program, found my first company, and never looked back.

Two years of apprenticeship, 13 projects shipped and defended, a SaaS architecture learned inside-out at a tech cooperative. Then I went freelance → first client in April 2023, consulting on dev and AI for startups and businesses.

Today I build agentic systems: agents, MCP servers, LLM pipelines, with Claude as my daily co-pilot. Three years shipping that independently, for 28 companies. Now I want to do it full-time, inside a team. Fast prototypes, no bullshit, proof over promises.

I've been the lost kid with no plan. That's exactly why I know how to structure chaos into something that works → for products and for people figuring out their path.

Let's talk

Thirty minutes, no filter. I answer everything, including the uncomfortable questions.

This page is for teams who are hiring. For consulting work, just email me directly.