Project
Build Feasibility

Build Feasibility

High for MVP — the bottleneck is data, not code.

The real bottleneck

The ingest layer is already built and published as five standalone libraries, so the remaining path to a first score is a collector that persists their output, a schema to persist into, and the scoring formula itself. Nothing on that list is blocked by an unanswered question.

The bottleneck is not code — it is threshold calibration. Two to three weeks of historical data is needed before scores can be trusted.

Ship the read-only scorecard first

Let it collect data while the rest is built. The clock on calibration starts the moment rows land in the database, which makes the collector — not the scoring engine — the next thing to build.