All work
AI InfrastructureOngoing

Aulë

A platform in progress, starting from the position that evaluation is not a feature of an AI system but the point of one.

Aulë

Overview

A platform built on one conviction: an AI capability should not be allowed to reach production until it has met a standard someone wrote down in advance. Quality as a property of the build, not a report written about it afterwards. In active development.

What I built

  • Wrote the formal architecture and technical vision as a corpus of research papers, currently unpublished.
  • Designed the system around a single rule: a capability that cannot demonstrate it still meets its standard does not ship.
  • Building the core services in Go and the evaluation layer in Python.
  • The public writing starts with the philosophy; the system itself follows when there is something to show rather than describe.