Work

Here is where that practice actually happens — the tools, run against real constraints, in the open.

Product01

Arcane Labs

Where the philosophy gets tested against something real — tools built to be interrogated, not admired, starting with federated learning.

Research02

SHARP-FL

The same test run as research: privacy-preserving systems for healthcare, built to survive scrutiny before they're trusted.

Research03

FedCore

A newer thread, still early: federated learning research built to be interrogated before it's trusted, wherever the data has to stay private.

Open Source04

ML Traceability Tools

Open-source tools for making machine learning experiments reproducible and auditable, so claims can be verified, not just believed.