Performance & Impact
Measurable latency, operational cadence, accuracy targets, and rider impact KPIs for Aegis.
Rider-first Impact
Aegis is designed so riders receive cash compensation as soon as the system detects a zone becoming unsafe or orders effectively disappearing.
Core KPIs
| Metric | Target |
|---|---|
| ML inference latency | < 80 ms per H3 zone |
| Zone state refresh | every 2 minutes |
| Zone intelligence aggregation | every 10 minutes |
| Redis feature hit rate | ~94% after warm-up |
| Dashboard/API latency (p95) | < 350 ms |
| H3 lookup | O(1) GPS -> zone mapping |
Risk scoring accuracy (Lf) | ~91% prototype gate |
| Premium pricing MAE | < INR 3.50 prototype gate |
| Fraud defense layers | multi-signal + hybrid gating |
How Success Is Measured
- loss ratio stays within planned margins (pool solvency)
- payout latency stays low during disruption events
- fraud false-positive rate does not block honest riders
KPI dashboard is represented by the KPI table above.