Rules Based Engine for Fintech - Credit Scoring
Budget: $10,000 – $20,000 USD
1 Requirements
1. Trust score engine will assign a trust score to a borrower based on signals received through app integrated with mobile SDK and external EKYC services. The trust score is used in the decision engine to decide borrowing capacity, amount, the interest charged etc. for the borrower.
2. The signals from App include SMS logs, call logs, contacts, social network activity, device/phone details, location, and EKYC details (if any). Data from EKYC services can be used based on the APIs and availability. Any previous history of lending in the platform can be used to further refine scores.
3. The engine will collect data coming from App/SDK, use APIs to pull data from external services, and parsers for various signal types, services, and docs. A Persona engine will extract user features/demographics/interests from various signals. Rules engine uses parsed signals (or features if deduction engine is used) and applies configured rules to assign a trust score to the borrower. Engines will work offline (once a day) to update scores. There is no requirement to process signals in real time.
4. Scores are exposed via API to the decision engine and other services.
2 Scope of Work
The scoring engine will be developed as an MVP and will be used as a stepping-stone for a production-ready engine. It will need a large amount of real data flowing from actual users to generate meaningful scores.
1. Design core architecture for a product-level implementation of the above functionality. The architecture will be scalable and extensible to sustain large-scale deployment 2. Define and implement core stack using proper components
3. Define signals to be captured by the SDK for SMS logs, call logs, device characteristics, social network activity (Facebook only).
4. Define APIs to send the events to score engine
5. Develop signal parsers for signals in (3)
6. Capture and store signals and attributes in DB
7. Develop a rules engine based on attributes found in (5) which determines the score. The score is updated every time new signals are detected from the borrower. Persona engine is NOT in the scope for MVP (requires ML and big data)
1. Trust score engine will assign a trust score to a borrower based on signals received through app integrated with mobile SDK and external EKYC services. The trust score is used in the decision engine to decide borrowing capacity, amount, the interest charged etc. for the borrower.
2. The signals from App include SMS logs, call logs, contacts, social network activity, device/phone details, location, and EKYC details (if any). Data from EKYC services can be used based on the APIs and availability. Any previous history of lending in the platform can be used to further refine scores.
3. The engine will collect data coming from App/SDK, use APIs to pull data from external services, and parsers for various signal types, services, and docs. A Persona engine will extract user features/demographics/interests from various signals. Rules engine uses parsed signals (or features if deduction engine is used) and applies configured rules to assign a trust score to the borrower. Engines will work offline (once a day) to update scores. There is no requirement to process signals in real time.
4. Scores are exposed via API to the decision engine and other services.
2 Scope of Work
The scoring engine will be developed as an MVP and will be used as a stepping-stone for a production-ready engine. It will need a large amount of real data flowing from actual users to generate meaningful scores.
1. Design core architecture for a product-level implementation of the above functionality. The architecture will be scalable and extensible to sustain large-scale deployment 2. Define and implement core stack using proper components
3. Define signals to be captured by the SDK for SMS logs, call logs, device characteristics, social network activity (Facebook only).
4. Define APIs to send the events to score engine
5. Develop signal parsers for signals in (3)
6. Capture and store signals and attributes in DB
7. Develop a rules engine based on attributes found in (5) which determines the score. The score is updated every time new signals are detected from the borrower. Persona engine is NOT in the scope for MVP (requires ML and big data)
Related categories:
Business, Accounting, Human Resources & Legal
Node.js
MongoDB
Microservices
Apache Spark