Data Sources

What are Data Sources?

Data Sources are external or internal systems that provide information for credit assessment and decision-making. TimveroOS integrates with these sources to retrieve data during workflows, which is then transformed through Mappings into Features for scoring and decision logic.

Data Source Types

External Data Sources

Third-party providers that deliver specialized information:

• Credit Bureaus (Experian, Equifax, TransUnion) - Credit reports and scores

• Open Banking (Tink, TrueLayer, Plaid) - Banking transactions and account data

• Identity Verification (Onfido, Jumio) - KYC and identity checks

• Employment Verification (Argyle, Truework) - Income and employment data

• Fraud Detection (Sift, Ekata) - Risk signals and fraud scores

• Valuation Services (Market Check, KBB) - Asset valuations

• Core Banking Systems - Internal customer data from existing systems

Internal Data Sources

System-generated data within TimveroOS:

• Application Data Source - Participant attributes from application forms

• Collateral Data Source - Asset attributes and characteristics

• Historical Data Source - Previous loan performance data

Data Source Architecture

Integration Principles

One Endpoint = One Data Source

• Multiple endpoints from the same provider require separate Data Source configurations

• Same endpoint used at different stages needs independent Data Sources

• Each Data Source has its own configuration, versioning, and Mappings

Example:

Experian Credit Bureau:
- Data Source 1: "Experian Preliminary" (soft pull endpoint)
- Data Source 2: "Experian Full Report" (hard pull endpoint)
- Data Source 3: "Experian Monitoring" (periodic update endpoint)

Role-Based Association

• Data Sources associate with specific participant roles (Borrower, Guarantor, Co-borrower)

• Data Sources can associate with specific asset types (Vehicle, Real Estate)

• This enables role-specific data retrieval

Data Flow

Workflow Execution
    ↓
Load Data Source Node (in Workflow)
    ↓
API Call to External System
    ↓
Raw Response Received
    ↓
Mapping Script Executes (JavaScript/Python)
    ↓
Features Extracted
    ↓
Save to Profile Node
    ↓
Features Stored in Participant/Asset Profile
    ↓
Decision Logic Uses Features

Configuration Process

Step 1: Framework Integration (SDK)

Technical setup by development team:

• API credentials configuration

• Endpoint specification

• Authentication setup (API keys, OAuth, etc.)

• Response format definition

• Timeout and retry logic

Step 2: Admin Panel Configuration

Business configuration:

• Test connection to Data Source

• View sample responses

• Create Mappings for data extraction

• Link Mappings to Features

• Associate Data Source with Workflows

Step 3: Workflow Integration

Connect Data Sources to decision flows:

• Add "Load Data Source" node to Workflow

• Select configured Data Source

• Add "Expression" node with Mapping

• Extract Features from response

• Add "Save to Profile" node to store Features

Data Source Configuration Patterns

Pattern 1: Tiered Data Strategy

Progressive data retrieval based on assessment stage:

Stage 1: Application (Low-cost, fast)

• Internal Application Data Source

• Basic identity checks

• Decision: Continue or decline

Stage 2: Assessment (Comprehensive, if Stage 1 passes)

• Credit bureau full report

• Open banking data

• Employment verification

• Decision: Approve, decline, or manual review

Stage 3: Final Verification (Pre-disbursement)

• Final identity check

• Fraud detection scan

• Final credit bureau check

• Decision: Disburse or hold

Pattern 2: Role-Specific Data Access

Different data for different participants:

Borrower:

• Full credit report

• Income verification

• Banking data

• Employment check

Guarantor:

• Credit report (lighter)

• Income verification

• Property ownership (if applicable)

Collateral Provider:

• Asset valuation

• Ownership verification

• Lien check

Pattern 3: Fallback Strategy

Redundancy for reliability:

Primary Source: Experian credit bureau Fallback 1: Alternative credit bureau (if primary fails) Fallback 2: Manual document upload (if all automated sources fail)

Mapping and Feature Extraction

Data Sources return raw JSON/XML responses that must be processed:

Mapping Purpose

• Extract specific values from responses

• Transform data formats

• Handle missing values

• Combine multiple data points

• Calculate derived values

Feature Creation

Mappings produce named Features:

• creditScore (from credit bureau response)

• monthlyIncome (from employment verification)

• debtToIncomeRatio (calculated from multiple sources)

• fraudRiskScore (from fraud detection service)

Profile Storage

Features accumulate in Participant/Asset Profiles:

• Profiles used by Offer Engine for pricing

• Profiles available to all Workflows

• Profiles persist throughout application lifecycle

Data Source Management

Versioning

• Each configuration change creates new version

• Previous versions preserved

• Rollback capability available

• Change tracking for audit

Monitoring

Track Data Source performance:

• Success/failure rates

• Response times

• Error types

• Cost per call

Error Handling

Configure behavior when Data Source fails:

• Retry logic (number of attempts, delay)

• Timeout settings

• Fallback options

• Manual intervention triggers

Common Data Sources

Credit Bureaus

Purpose: Credit reports, scores, identity verification Typical Mappings:

• Credit score extraction

• Trade line analysis

• Inquiry count

• Derogatory items

Integration Consideration: Soft vs hard pulls, cost per query

Open Banking

Purpose: Transaction history, account balances, income verification Typical Mappings:

• Average monthly income

• Recurring payments identification

• Account balance trends

• Overdraft frequency

Integration Consideration: User consent required, real-time access

Employment Verification

Purpose: Income, employment status, tenure Typical Mappings:

• Current salary

• Employment length

• Position title

• Employer name

Integration Consideration: May require payroll system integration

Fraud Detection

Purpose: Risk signals, device fingerprinting, behavioral analysis Typical Mappings:

• Fraud risk score

• Device reputation

• Velocity checks (multiple applications)

• Email/phone validation

Integration Consideration: Real-time scoring, IP geolocation

Implementation Resources

Through Admin Panel (Step 2)

Data Source Setup:

• Data Integration - Overview

• Data Sources - Configuration

• Feature Store - Feature management

• Metrics Engine - Calculated metrics

Through SDK (Step 1)

Framework Integration:

• DataSource Integration - Technical setup guide

Related Topics

• Mappings - Data transformation scripts

• Workflow Tool - How Data Sources integrate with workflows

• Decision Configuration - Complete decision automation overview

---

TimveroOS: Integrated data access for informed credit decisions