Please sign up or log in to continue.

QShield - Parametric Income Protection for Gig Workers

QShield automatically compensates Qcommerce delivery partners for income lost to rain, extreme heat, poor AQI, curfews, and platform outages no claims, no waiting, just instant parametric payouts.

Comment

Inspiration

Every monsoon season, millions of Q-commerce delivery partners in India lose entire days of income to rain, heat, or platform outages — with zero recourse. A Zepto rider earning ₹5,600/week loses ₹800 on a single rainy day. No insurance product existed specifically for this. We built QShield to fix that.

What it does

QShield is a weekly parametric income protection system for gig delivery workers on Zepto, Blinkit, and Swiggy Instamart. Workers pay ₹168/week (platform covers ₹112 more). If a verified disruption day occurs — heavy rain, extreme heat, severe AQI, a government curfew, or a platform outage — the system automatically validates it against real-time APIs and processes a payout. No claim forms. No calls. No waiting.

The premium is priced at 5% of weekly income — the industry benchmark for accessible parametric coverage. Each disruption day pays out 50% of the worker's daily baseline income, capped at 2 days per week.

How we built it

  • Backend: FastAPI with a premium calculation engine, parametric trigger service, and dual-validation system — environmental threshold must be crossed AND zone order volume must drop ≥25% before a disruption day is confirmed
  • Trigger APIs: OpenWeatherMap (rain, heat), AQICN (real-time AQI), NewsAPI (curfew/strike detection), UptimeRobot (platform outage monitoring)
  • Dark store data: 4,081 real Zepto, Blinkit, and Swiggy Instamart store locations scraped from public APIs — used for zone-level delivery mapping
  • Frontend: React + Vite with separate worker and admin dashboards
  • Fraud detection: Multi-signal adversarial defense system — GPS spoofing is irrelevant by design since triggers are zone-level not worker-level

Challenges we ran into

OpenAQ's v3 API returned no data for Indian cities — we switched to AQICN which gave clean real-time AQI in a single call. Designing the dual-validation trigger logic to be both fraud-resistant and fair to genuine workers was harder than expected. Our original premium formula produced ₹1,344/week — no gig worker would pay that. We rethought the entire model and anchored it to the 3–5% of weekly income industry benchmark instead.

Accomplishments we're proud of

Real parametric trigger logic connected to 4 live APIs. A fraud-resistant architecture where basic GPS spoofing is irrelevant by design — triggers are zone-level, not worker GPS-based. A financially viable premium that a real ₹5,600/week worker can actually afford at ₹24/day.

What we learned

Parametric insurance math is harder than it looks — loss ratios, payout caps, and premium benchmarks all have to be consistent with each other or the product is not viable. Real-world APIs behave very differently from their documentation. And spending time on product logic before writing code saves enormous amounts of time later.

What's next

Phase 2: live worker registration, PostgreSQL database, Razorpay sandbox payouts, dynamic premium engine. Phase 3: trained ML model for disruption probability, Z-score fraud detection, mobile-responsive UI, pilot with real delivery partners in Bengaluru.

Built With

Share this project:

Updates