Nicolas Bonnefoy

Nicolas Bonnefoy

CTO, deploying AI systems into production
I deploy AI into production, alongside your teams.

Summary

CTO focused on taking AI systems into production, with twenty years of experience across gaming, AdTech, EdTech, FinTech, web3 and AI. Most recently, he was CTO of a generative AI platform in production for enterprise clients, where he designed the multi-agent architecture and ran operations. He has also led pre-sales on client projects, including the winning NFT tender for France’s Ligue Nationale de Rugby (LNR), and data compliance work (GDPR on student data, security audits).

Core skills

AI in production
Multi-agent orchestration • Temporal • LLM red teaming • Generation harness • Model registry
Pre-sales and client work
Tenders • Technical demos • AI project scoping • Build vs buy • Due diligence
Compliance and data
GDPR • Minors’ data • GAR accreditation • Hosting in France • Security audit
Engineering leadership
Team leadership • Legacy takeover • Zero-downtime migrations • Cloud cost control
Product and R&D
Idea to launch • Rapid prototyping • UX design • Measure and iterate
Domain expertise
Generative AI • EdTech • Gaming • AdTech • Web3 • FinTech

Technical expertise

AI: Multi-model orchestration (OpenRouter, Runware, Gemini, Mistral), LangGraph, Temporal, structured outputs (JSON Schema), evaluation and red teaming of generative models, reproducible generation harness, LoRA, coding agents (Claude Code)
Database: PostgreSQL
Languages: TypeScript, JavaScript, Python, Solidity
Frameworks: Vue.js, Node.js
Infrastructure: Docker, CI/CD, Google Cloud Platform, Firebase
Blockchain: Ethereum, Polygon, Solana, Seaport, smart contracts

Featured projects

Generative AI platform in production, confidential project
Generative AI platform in production

CTO of a B2B generative AI SaaS platform in production, for enterprise clients.

See details →
Et voilà, a virtual business card generated from a resume
Et voilà (etvoila.cv)

Free virtual business card generated by an LLM from a resume, for job seekers with limited digital skills.

99%
fields correct
2 weeks
from spec to beta
100%
hosted in France
See details →
OVAL3, rugby player cards
OVAL3

Won the NFT tender from the Ligue Nationale de Rugby (LNR), France’s professional rugby league, then launched a real-time on-chain auction platform.

LNR
tender won
$15,000+
from the first sale
$60k to $100k
monthly revenue
See details →
Madmagz, collaborative magazine creation platform
Madmagz

Takeover and rebuild of collaborative magazine creation software more than ten years old, used by schools and companies.

2M+
registered users
25 TB
migrated with no downtime
GAR
Ministry of Education accreditation
See details →

Professional experience

NIOBIUM 41since Oct. 2025

Founder
  • AI consulting and deployment: scoping projects before budget is committed, build vs buy decisions, technical audit, taking systems into production with the client’s teams.
CTO of an AI SaaS platform (confidential)Nov. 2025 to Nov. 2026
  • Designed the architecture, made the technical choices, led development and ran operations.
  • Multi-agent system in production for enterprise clients.
  • Infrastructure overhaul, with a zero-downtime production cutover.

ERA2140, web3 studioAug. 2020 to Oct. 2025

Managing Partner / CTO
  • Studio profitable from its first year and self-funded, more than one million euros in annual revenue.
  • More than 40 projects shipped, on Ethereum, Polygon and Solana.
  • Around ten employees, up to 18 people including freelancers.
  • Clients including OVAL3 (rugby), Fungiball (tennis) and Darewise (video games).
  • Infrastructure on Google Cloud Platform for blockchain applications.

OVAL3Jul. 2022 to Jul. 2024

Board Member / CTO
  • NFT tender from the Ligue Nationale de Rugby (LNR), France’s professional rugby league, won in October 2022 (rights for the players of the 30 Top 14 and Pro D2 clubs): owned the technical, compliance, infrastructure and CSR sections of the bid.
  • Real-time on-chain auctions proposed by Nicolas, first turned down by the client, then adopted after a technical demo.
  • First official auctions on 21 and 22 December 2022: 31 cards, more than $15,000. The platform then generated $60k to $100k in monthly revenue.
  • License extended to Major League Rugby (9-year contract).

The OutpostMar. 2023 to Dec. 2024

Co-founder
  • Venue in Paris for international web3 teams, with an events program.

Madmagz, SaaS magazine platformApr. 2017 to Aug. 2020

CTO / CPO
  • Took over software more than ten years old: stabilization, then a rebuild of the platform.
  • More than 2 million registered users, including 1 million active users.
  • Migrated 25 TB of data to a cloud architecture, with no downtime for clients.
  • GDPR compliance for student data (anonymization) and French Ministry of Education GAR accreditation.

Video game e-commerce platform (private sales)2016 to 2017

Founder
  • Development, product design and go-to-market.
  • 10,000 sales within 60 days of launch.

On-demand storage startup2016

Co-founder
  • Built the technical platform, from idea to launch.

Prodigious (Publicis Groupe)2015 to 2016

Technical project manager
  • Moved the production teams from Flash to HTML5 in two months.
  • Projects for major accounts: Renault, Nissan, SNCF, Euro 2016, Netatmo.
  • Modular email templates: integration time cut from 4 to 5 days down to half a day.

Mediastay2010 to 2015

Lead developer, then head of the game studio
  • Set up an in-house game studio working on TV licenses: “Qui veut gagner des millions ?” (Who Wants to Be a Millionaire), “Les 12 Coups de Midi”.
  • Templatable game engine delivered in one week, replacing outsourcing at €15k per game.

Education and certifications

Professional training • */ut7*2011 to 2012

  • Scrum Master certification
  • Agile methodologies
  • Advanced product management

Approach

Tech for business. An AI project is judged in production, not in a demo. I start by scoping the need and pricing the options before any budget is committed. Then I decide on measurements: models compared on test sets, real running costs. And I stay through to production, with the teams who will run the system. What I care about is a product that solves a real problem and makes money.

Languages

  • French: Native proficiency
  • English: Full professional proficiency
  • Spanish: School level
Open to a role in applied AI, or to deployment engagements through NIOBIUM 41.