Overview
A capability passport for AI fluency: assessments of how a person works with AI, based on observed collaboration.
Problem
Most “AI skill” scores count activity. That measures usage, not effectiveness, and it rewards the wrong evidence.
My Role
I designed the evaluation protocol and implemented ingest, a privacy guard, persistence and the private workspace.
What I Built
- A versioned evaluation protocol that refuses to score prompt, token or tool counts.
- Ingest for protocol fetch and assessment submit.
- A privacy guard that rejects raw emails, URLs, paths and code fences.
- Persisted assessments and an authenticated workspace.
Architecture & decisions
A protocol-first backend: ingest and persist assessments, then aggregate them into a private capability view. Scoring is based on observed collaboration, not usage counts.
Technologies
- Spring Boot
- Angular
- PostgreSQL
- Flyway
Outcome
A working prototype of the protocol, ingest path and private workspace.