Zero to enterprise-ready in two months. Multiple funding rounds followed.

An institutional analytics platform that scores bond issuance and stock debt health via a proprietary rating algorithm. Banks, asset managers, and enterprises use it to price corporate debt and run ex-post analysis against realised market outcomes. Multiple funding rounds closed.

Two months, zero codebase, banks on the other end of the demo. The vendor API was out of budget, so the data path ran on quarterly Excel exports, manually downloaded, format-drifting, no API safety net. The proprietary rating logic had to be defensible enough for institutional buyers to stake decisions on, and ex-post analysis had to be first-class because that's what they actually buy.

Constraint: institutional analytics on a startup's licence budget, with Excel as the only realistic source. The non-obvious call: design the data path around idempotent reruns and audit-grade ex-post replay, not streaming freshness, that's what banks actually use. Concretely: Excel ingestion with row-level schema validation that catches vendor format drift before it corrupts ratings, the proprietary bond and stock debt-health engines, NestJS API, React dashboard, JWT auth, Postgres schema built around quarterly snapshots. Test-driven end to end. The proprietary algorithms were complex enough that without the suite the platform wouldn't have been buildable, every new requirement could quietly break three others. What it bought: institutional buyers who could replay every rating call against realised outcomes.