Deterministic replay · tamper-evident decision logs · multi-tenant RLS · economic optimization.
- Deterministic replay (snapshot + model version bound decisions)
- SQL-first core logic in PostgreSQL
- SHA256 hash-chained decision ledger
- Economic objective (revenue / fraud / capital / churn)
- Multi-tenant isolation (RLS)
| Metric | Target |
|---|---|
| Throughput | ≥ 5k TPS sustained |
| Decision latency | p95 < 200ms |
| Fraud latency | p99 < 300ms |
| Replay | 30 days in < 4h |
Event inbox → validation/idempotency → domain streams
→ features → risk score → policy → economic eval
→ decision ledger (hash chain)
→ observability / replay / stress
Author: Felipe Fernandes · Orbit
Try it in the browser (no clone): see badge / homepage above, or the Labs hub.
Portfolio: felipeofdev-ai.github.io · Author: Felipe Fernandes · felipe.of.dev@gmail.com
If this helped you, star this repo — organic only. No bots, no paid stars.