My name is Pierre Chaumont, an AI Engineer based in Bordeaux. I design and ship local, open-source and self-hosted solutions that pair technical robustness with real business value. I build document extraction APIs, chatbots wired to internal documentation (RAG), tool-using agents that drive services in plain language, and large-scale data processing pipelines.
A Python library that keeps personal data out of a language model without breaking the application around it. PII is detected (regex, NER or another LLM) and replaced with stable placeholders, so that john.doe@example.com becomes <<EMAIL:1>>. The model only ever works on de-identified text, and the real values are restored in the response, so the end user never sees the de-identification. A tool called by the agent receives the real value, while the LLM deciding to call it still sees only the placeholder. Around the library sit an inference API, a hub, a demo application and a design system.
A secure, production-ready API key system, agnostic of the framework around it. Pluggable hashing strategies (Argon2, bcrypt or your own), backend-agnostic persistence (SQLAlchemy, in-memory, or your own repository) and an optional cache layer (aiocache, Redis). Ships integrations for FastAPI, Litestar, Django and Quart, plus a Typer CLI to manage keys. Formerly fastapi-api-key.
A library that makes an LLM play chess, taking advantage of the fact that a model can reproduce the PGN format of a game. It provides the tooling to run experiments, edit the PGN during inference and instrument games. It started as a way to check whether ChatGPT really holds an 80-move game together without an illegal move.
A chatbot for talking to PDFs through a RAG system that hands the model only the relevant chunks of the document. A chat memory is built in, which allows long discussions with a single document.
A fishing bot for World of Warcraft, fully automatic. Mouse clicks and movements happen inside the game with no human intervention, and a web page lets you watch live what the bot is doing.
Conway's Game of Life, rendered in real time with Pygame. A grid of cells whose birth and death rules produce gliders, oscillators and still lifes out of a handful of lines of code.
Python
Pygame
Testimonials
What it is like to work with me.
End-of-engagement reviews and recommendations from people I have worked with.
End-of-engagement review
I highly recommend Pierre for any opportunity in Data and Artificial Intelligence. Pierre has shown exceptional skill in these areas, pairing a solid theoretical grasp with remarkable practical application. He has also consistently shown great professionalism, an excellent work ethic and a constant willingness to learn and improve.
I got to work alongside my colleague on various open source projects, and it was a great experience. Pierre has a Data Scientist profile with a real appetite for software engineering, handling Git, CI/CD, Docker and the cloud. His rigour in applying development practices is what makes the work last. Add a good collaborative spirit and a real dedication to open source.
Gabriel P., Data Scientist · Nov 21, 2024 · Malt
Timeline
My life changelog.
The path, the diplomas, and a few titles along the way.
2025
AI Engineer at FIDUCIAL
Experience
Jan 2025 to present · 1 year 9 months
Design of an API for processing unstructured documents, built for structured extraction from notarial paperwork, with parsing, classification and extraction. It runs in production at around 20,000 requests a month on deeds of sale, ID cards, questionnaires and the other pieces of a case file. Industrialisation of a shared RAG foundation, turned into business chatbots for the group's notaries, garages, pharmacies and its other lines of work. Design of an LLM orchestration API running on local models, with context handling and persistent memory, and of tool-using agents that drive services and data components in plain language, such as Matomo or Supabase. Plus the observability platform that goes with all of it.
2024
"AI-Powered Software and System Design" certification
Certification
A DeepLearning.AI certification covering AI-assisted software architecture design, building and optimising databases with the help of LLMs, and applying advanced design patterns.
Run alongside the Scalian and then FIDUCIAL positions, with the employer's agreement and outside working hours. Building data pipelines and AI tooling, with quality follow-up and delivery on short timelines.
Data Scientist at Scalian
Experience
Nov 2023 to May 2024 · 7 months
Within a five-person data team, design and industrialisation of an OCR pipeline extracting structured data from millions of handwritten cadastral plans, owning the move to production. Development of several AI-driven systems deployed on cloud infrastructure, among them a GitLab project analysis chatbot, automated unit test generation and urban pollution simulation.
2021
Data Scientist at Arkema, apprenticeship
Experience
Sep 2021 to Sep 2023 · 2 years 1 month
In the purchasing department, predictive machine learning models for supplier price evolution, wired to Power BI dashboards. A Python library automating SAP ETL workflows behind a graphical interface. And a dynamic web questionnaire application integrated with Microsoft Teams and driven by interaction data.
Master's degree in AI & Big Data, ESGI
Education
Sep 2021 to Jun 2023 · 1 year 10 months
Deep learning, natural language processing, deep reinforcement learning, BI and Big Data, Spark and functional programming, cloud and automation for machine learning, IT project management and agile methodology.
Web developer at Université Claude Bernard Lyon 1
Experience
Feb 2021 to Apr 2021 · 3 months
Development of a web application for ticket management and for booking time slots on a spectrometer.
2020
Bachelor's degree in software architecture & project management, ESGI
Education
Sep 2020 to Jun 2021 · 10 months
Python scripting, advanced C and Java, Node.js APIs, Angular front ends, unit testing, advanced algorithms, relational database design, artificial intelligence, Linux and agile project management.
2019
Junior runner-up, Regional Youth Championship
Award
Échecs, C.R.M.L.E. The game stuck around, since the Bresse project makes an LLM play chess and chess_pytorch_ai tries to get a neural network to do the same.
Junior runner-up, ARA League Championship
Award
Chess, ARA League.
2018
Associate's degree in IT services for organisations
Education
Foundations in software development, networking and information systems management.
