The requirement management tool the safety case was missing
Client & context
Inside a live safety engagement at AstaZero (RISE), the Test Robot safety case needed requirement traceability and evidence management that existing tooling could not provide — and procuring an enterprise requirements platform was neither fast nor a good fit.
The problem
Safety work dies in spreadsheets: requirements without traceability, evidence scattered across drives, and no audit trail for what changed and when. The safety argument for SAE L3-L4 test equipment demanded traceable, immutable, reviewable requirement data — and it didn't exist.
What I did
- Designed and shipped, solo and AI-assisted, a full-stack Requirement Management web app used directly in the safety process.
- Backend on Fastify 5 with PostgreSQL 16: 19 tables, append-only history, and immutability triggers so every requirement change is a permanent, auditable record.
- JWE session authentication (@auth/core + jose) and file uploads for evidence attachments.
- Vanilla-JS SPA frontend with a REST-native domain layer — zero framework overhead, fast and maintainable.
- Encoded the standard into the data model: traceability, immutability, and evidence as first-class citizens, not just CRUD screens.
The outcome
A requirement-traceability tool the safety process now runs on, delivered end-to-end by one engineer inside the safety engagement — proof that AI-assisted development changes what a safety consultant can deliver.
Standards & tools
ISO 26262ISO/TS 5083:2025Fastify 5PostgreSQL 16JWE authAI-assisted development
Facing something similar?
This is the kind of work I do under Safety Tooling & Simulation, Built to Order.
Book an intro call