Protocol Engineer · PhD in AI · Paris, remote
I build the systems DeFi protocols run on (indexers, APIs, trading venues) and the AI agents that quote, negotiate and trade on them. I measure what I build, and myself.
Years shipping software
12
since 2014, onchain since 2022
Autonomous trading agents
~20
from Safe smart accounts
Chains indexed live
22
1.3M liquidations, 36 lending protocols
GitHub contributions
3,976
1,712 in the last 12 months
Peer-reviewed papers
13
AAAI, AAMAS, PhD in AI
Protocol EngineerCork
2025 - now
Tech LeadNefture
2024 - 2025
Senior Data EngineerOrange Business
2019 - 2024
Research EngineerUniversity of Montpellier
2018 - 2019
PhD Researcher and LecturerUniversity of Montpellier
2015 - 2018
Junior Full-Stack DeveloperOpen Management Systems
2014 - 2015
2 of 4 goals met in October
Weekly counts, last 26 weeks. Code from GitHub, the rest from Ananke and Mnemo.
+18 building
Every day gets a rating: a good one adds, a bad one subtracts, capped at ±30. The line is where the recent days leave me. From Ananke; gaps are days without a rating.
Liquidation research and bots ↗
Indexed over 1.3 million lending liquidations and $9.7B repaid across 36 protocols on 22 chains (EVM, Solana, Tron), 2019 to the current block, to measure who wins and why.
A Rust fork of the Zed editor for running many AI coding agents at once, with a live registry across windows and a status bar that jumps to the agent needing attention.
My own apps on my own infrastructure: Mnemo, a todo app that draws projects as galaxies with an API for AI assistants; Ananke, life objectives and habits; Phorcys, training.
Security research
Found a signature-validation flaw in a widely used smart-account module.
Defeasible reasoning for existential rules
PhD thesis, 2018: what a system should conclude when its data disagrees. First-author papers at AAAI and AAMAS.
Google Scholar →300+ hours of university teaching.
Remote, EU and US hours. English, French, Arabic.