I’m building Kepler with John McRaven. We’re building the AI platform for financial services, where every answer traces back to its source filing, page, and line item.
We’re a team of engineers who’ve built mission-critical systems at Palantir, Citadel, and PhysicsX. We’re backed by some of the architects of modern data and AI platforms, including founders of OpenAI, Facebook AI, MotherDuck, dbt, Outerbounds, and others.
Email me about Kepler. • Follow along on the Kepler website or LinkedIn page.
Where my time goes
- Kepler. Most of it. Recruiting, customer conversations, and shipping the platform with the team.
- Writing. When something is hard enough to explain that I want to write it down, it ends up on this site.
- Teaching. Courses on LinkedIn Learning. My latest is Fundamentals of AI Engineering.
- Advising. A small number of founders building data and AI infrastructure.
What I’m thinking about
- Why verifiable AI is the only kind of AI that will work in regulated industries
- How Forward Deployed Engineering scales, and where it breaks
- Context engineering as a data engineering problem in disguise
- The tooling gap between AI demos and AI in production
Recently
- How Kepler built verifiable AI for financial services with Claude: published on the Anthropic customer blog
- Build Products Like a Forward Deployed Engineer: talk on the FDE mindset for product builders
- The Definitive Guide to Forward Deployed Engineering: the role, the skills, and how Palantir’s Project Frontline scaled it
- Context Is The Easy Part: why context engineering is really an engineering problem
Read more about me or check out my writing, talks & courses, and projects.