Work with me
I'm a senior data scientist at Dataiku. I spend my days getting AI systems into production for clients in regulated industries: pharma, fintech, insurance, energy. Before that, seven years at SLB doing production ML on oilfield equipment.
Outside of that, a few things I say yes to.
Speaking
I talk about what actually breaks when you put LLM agents in front of real industrial data. Not the demo version. Topics I know well enough to take questions on:
- Agentic systems in regulated environments, and the audit trail problem
- Retrieval over messy proprietary formats, where the parsing is the hard part
- What production ML in energy taught me that transfers to GenAI, and what does not
- MCP as an integration layer for systems that predate the internet
Podcast guests
I host Software Engineering Radio for the IEEE Computer Society, roughly five episodes a year. I look for guests who have shipped something hard and can explain the parts that went wrong. If that is you, or you know who it is, tell me. Pitches from people describing their own work are welcome and I read all of them.
Advisory
Occasional short engagements:
- Technical review of an AI system before it goes in front of regulators
- Second opinion on an agent architecture that is not converging
- Expert commentary for press or research
I am employed full time, so anything here is subject to my employer's conflict of interest policy. If there is overlap I will say so early rather than late.
What I don't take
Being clear about this saves us both a call.
- Full-time or contract implementation work. I have a job I like.
- “Can you build us an MVP.” No.
- Unpaid technical due diligence for investors.
- Anything that needs me to pretend a model is more reliable than it is.
Start a conversation
Fifteen minutes, no agenda required. If it turns into something bigger we can figure that out after.