Most enterprise AI efforts fail for the same reason: the model gets selected before the workflow is understood, or the agent gets deployed before the process is actually fixed.

I've spent fifteen years learning to do it in the right order — observe the work, structure the data, build the foundation, then add intelligence where it compounds.

Transformation work needs discipline. It also needs curiosity, drive, and the entrepreneurial reflex to spot the opportunity that hasn't been seen yet.

Currently leading innovation and automation for Operational Risk at Société Générale CIB. With my team, we're on track to automate 80% of RISK NFR monitoring, control, and reporting day-to-day — Streamlit-first, pipeline-up, knowledge-graph-backed — and the GenAI layer now going on top sits on something that actually works.

Before SocGen: digital transformation of Americas Finance at BNP Paribas CIB (where I introduced the first ML application into the Finance department), aftermarket analytics at Pratt & Whitney Canada, and revenue management at Hertz across the French Caribbean.

I build outside work because it keeps me current on the stack and because I'm a builder by nature. DOMn.fr is the long-running one — rebuilt in 2025 around AI-driven content gathering at ~6,000-business scale. I'm also building an event-driven macro intelligence platform that pairs LLM event classification with structured market data.