I'm Florian Metzger-Noel, a computer scientist with a particular interest in distributed systems, knowledge graphs and automation with compositions of LLMs.

I've been programming since 1990, when learning to code meant working through printed reference books and writing BASIC on a monochrome 8086. I got hooked and never stopped building after that.

At 21 I co-founded a software company, grew it over eight years and finally sold it. While running the company I studied fine art at Berlin University of the Arts, working with generative art and genetic algorithms.

I played jazz double bass professionally for 15 years, more than 300 concerts across 34 countries. I co-founded a cultural association, ran a fringe theatre for two years, and built a restaurant, which I grew and managed over nine years. Parallel to running the restaurant I studied for a computer-science degree.

Since 2022 I've worked as a software architect at ML!PA, where I started what is now L-IoT — a platform for managing industrial device fleets. The idea existed in fragments, prototype scripts and notes. Those I formalised into an architecture, prototyped it, and built the team that turned it into what is today the company's flagship platform used by Knorr-Bremse, KAEFER and Liebherr, among others.

Today my work has shifted into open source research published under my name, carried inside my contract at ML!PA. I focus full time on Ranke-Graph, a provenance-first data structure for verifiable knowledge graphs: machine-readable institutional memory in which every answer traces back to its sources. Two papers and the specification are written, with a reference implementation in Go. Alongside it I build Sindri, an open-source sandboxed orchestrator for AI coding agents that makes them cooperate like members of an organisation. I've worked with LLMs since the GPT-3 era and have run production systems on them since 2023, in educational and industrial use cases.

I speak on both halves of that work: the industrial edge, and the research. Both live where disciplines meet, which is usually where the interesting problems are.