Not authentication.
A decision.
Same surface — “user behavior.” Different problem, different signals, different output.
| Feature | D1 (US11101993B1) | Present invention |
|---|---|---|
| Core focus | Authentication & authorization | Dynamic restriction of payment options at checkout |
| Data type used | Biometric & behavior signals | Transactional metrics — POD %, rejections, returns, demographics |
| Stage of application | Before login / system access | At e-commerce checkout, in real time |
| Output | Access granted / denied | Restriction or allowance of POD and other payment options |
| Technical effect | Secure access | <2s decision latency, >95% fraud accuracy, 40% fewer POD failures |
Implemented on distributed servers, in-memory stream processing, and an event-driven API layer — not a software algorithm in the abstract.
A system for dynamically managing payment options within an e-commerce platform, comprising:
- 01
Data collection module gathering user interaction data from transaction databases, user accounts and external APIs — order details, payment methods, rejections, returns, demographics.
- 02
Integration module consolidating data into a centralized lakehouse for real-time access.
- 03
Preprocessing and feature-engineering module computing behavioral metrics including POD percentage and rejected/returned POD orders per user.
- 04
Predictive ML model trained on historical data to classify high-risk users, applying thresholds such as POD rejection > 30% in last 3 months.
- 05
Real-time decision engine on distributed in-memory servers, evaluating eligibility within < 2 s and dynamically restricting payment options based on risk metrics and ML predictions.
- 06
Application integration layer of APIs and microservices enabling secure, event-driven communication between backend engine and checkout frontend.
- 07
Monitoring and optimization tools — anomaly detection, decision-accuracy, fraud-rate, latency and satisfaction telemetry — with iterative refinement.