Tech Stack & Data
Storage, external signals, and data-quality controls powering zone-time decisions and auditable payouts.
Core Data Stores
| Store | Role |
|---|---|
| PostgreSQL | transactional data: profiles, policies, contracts |
| TimescaleDB | time-series triggers and audit histories |
| MongoDB | worker activity logs and claim records |
| Redis (feature store) | low-latency online features for Lf, Ew, zone context |
| S3 data lake | historical datasets + reproducible ML training artifacts |
External Data Dependencies (Signals)
- IMD: rain, flood/heat-related signals
- OpenAQ / CPCB: AQI and pollutant severity patterns
- NDMA: disaster declarations and curfew alerts
- Maps / routing signals: route constraints and mobility disruption context
- Platform API (simulated): zone halt + demand/drop indicators
- Uber H3: hexagonal indexing for micro-zone eligibility
Data Flow Model (Minimal)
Raw signals -> normalization -> zone/time event schema -> time-series persistence -> online features -> trigger engine + AI scoring.
Data Quality Controls
Aegis keeps decisions auditable by enforcing:
- timestamp normalization (avoid ordering drift)
- zone ID canonicalization (deterministic eligibility)
- missing value handling before scoring
- immutable claim/audit records for payout explainability