Nexora OS: An AI-Powered Income Stability System for Gig Workers

Introduction

Gig economy workers rely heavily on daily earnings, making them highly vulnerable to disruptions such as extreme weather conditions, poor air quality, and regional shutdowns. Traditional insurance systems are often inefficient in addressing these challenges due to manual claims processes and delayed payouts.

Nexora OS is designed as an AI-powered income stability system that automates disruption detection, claim generation, fraud verification, and payout processing. The objective is to ensure timely and reliable financial protection without requiring manual intervention from users.

Overview

Phase 2 focused on transforming the initial concept into a fully functional, end-to-end system. The implementation emphasizes real-time automation, system-level integration, and AI-driven decision-making.

The platform operates as a continuous monitoring system that evaluates environmental conditions and triggers appropriate actions when predefined thresholds are exceeded.

System Architecture

The system follows a modular microservice-based architecture to ensure scalability and separation of concerns.

Technology Stack

Frontend:

• React 18 (Vite)


• Tailwind CSS


• Component-based architecture using shadcn/ui

Backend:

• Node.js (Express)


• RESTful APIs


• Scheduled task execution using node-cron

AI/ML Layer:

• Python (Flask microservice)


• scikit-learn models

Database and Authentication:

• Supabase (PostgreSQL with integrated authentication)

External Integrations:

• OpenWeatherMap API for weather data


• OpenAQ API for air quality data


• Razorpay sandbox for payment simulation

Deployment:

• Frontend hosted on Netlify


• Backend and AI services hosted on Render

End-to-End Workflow

The system implements a complete automated pipeline:

User → Risk Analysis → Policy → Trigger Detection → Claim → Fraud Check → Payout

This workflow ensures that all operations are handled automatically without requiring user-initiated actions.

User Authentication and Onboarding

The platform includes a structured onboarding process:

• Multi-step user registration


• Email and password authentication


• OTP-based verification

The system collects the following data:

• Personal details


• Platform information (e.g., Zomato, Swiggy)


• Weekly working hours


• Location


• UPI ID

This data is used to personalize risk assessment and premium calculation.

AI-Based Risk Profiling and Premium Calculation

A dedicated Python microservice calculates personalized premiums based on user-specific parameters.

Model details:

• Algorithm: Linear Regression


• 
  

  Inputs:

  
  
  
  
  
  • Location-based risk score
  
  
  • Weekly working hours
  
  
  • Platform type
  
  
  • Claim history
  
  
  
  
  
  



• 
  
  

  Output:

  
  
  
  
  
  • Weekly premium within a defined range
  
  
  
  
  
  

The backend communicates with the AI service through APIs to dynamically determine pricing.

Policy Management System

The policy framework is based on a parametric model with predefined thresholds and payouts.

Example coverage structure:

• Heavy Rain (>20mm/hr): ₹400


• Extreme Heat (>43°C): ₹300


• Hazardous AQI (>300): ₹350


• Flood Alert (Government Red Alert): ₹500


• Bandh/Curfew (Admin-triggered): ₹450

Policies are stored and managed within the database and linked to user profiles.

Real-Time Trigger Detection

A scheduled process runs at regular intervals to:

• Fetch environmental data from external APIs


• Evaluate conditions against predefined thresholds

When conditions meet trigger criteria, the system automatically initiates the claim process.

Automated Claim Generation

The system eliminates manual claim filing by:

• Automatically generating claims upon trigger detection


• Recording claim details such as event ID, timestamp, and trigger type

This ensures a seamless and immediate response to disruptions.

Fraud Detection System

Fraud detection is implemented using a machine learning-based approach.

Model:

• Isolation Forest

Validation checks include:

• GPS location consistency


• Weather data verification


• Time-based validation


• Behavioral pattern analysis


• Duplicate claim detection

Each claim is assigned a fraud score. Claims below a defined threshold are approved automatically, while others are flagged for review.

Payout System

Once a claim is verified:

• The payout process is triggered automatically


• Payment is simulated using the Razorpay sandbox

The system records:

• Payout amount


• Processing time


• Transaction reference

The entire process is completed within a short time frame, demonstrating real-time capability.

Worker Dashboard

The user interface provides a comprehensive dashboard displaying:

• Active policy status


• Claims history


• Total protected income


• Real-time environmental data including rainfall, temperature, and air quality

This enables users to understand their coverage and system activity.

Admin Dashboard

A separate administrative interface provides control and monitoring capabilities:

• Policy management


• Claim tracking


• Fraud analysis


• Revenue overview

Administrators can also manually flag disruption zones and trigger payouts when necessary.

Data Flow and Integration

The system ensures seamless communication across components:

• Frontend communicates with backend through APIs


• Backend interacts with the database for storage


• Backend integrates with AI services for risk and fraud analysis


• External APIs provide real-time environmental data

Deployment

The application is deployed using a cloud-based approach:

• Frontend on Netlify


• Backend and AI services on Render


• Database and authentication via Supabase

Continuous deployment is enabled through version control integration.

Improvements from Phase 1

Phase 2 introduced significant advancements:

• Transition from prototype to functional system


• Complete automation of the claim lifecycle


• Integration of real-world data sources


• Implementation of AI-driven models


• Structured user journey and workflows


• Dedicated administrative controls

Conclusion

Nexora OS represents a shift from traditional insurance systems to an automated, real-time income protection platform.

By integrating AI, real-time monitoring, and automated execution, the system delivers a scalable and efficient solution for gig workers.

The project demonstrates a strong emphasis on system design, automation, and practical applicability, aligning with the objectives of building impactful and intelligent solutions.

Team SyncShift