Data Vault
- “Single source of truth” for analytics
- Systematically automate approval process with strong auditability
- Develop sophisticated data aggregates easily
- Manage lineage and permissible use tracking dynamically
Feature Engineering
- Register sophisticated features using other features, data elements and aggregates
- Dynamically manage lineage and permissible use and catch exceptions
- Make approved features available using API’s to modelling teams
Model Studio
- Insert high power AI models into the decision workflow with full traceability
- Standardize and automate model approval process that meets regulatory requirements
- Perform extensive experimentations
- Enable continuous model validation tracking
Products & Frameworks
- Develop real-life product hierarchies and parameters for products
- Develop complex frameworks for decision optimization like NPV, CLTV, Loss Estimates
- Facilitate decisions at “segment-of-1” using tightly integrated frameworks within policies
Policy Development (use cases in Prospecting, Underwriting, Customer Mgmt. and Fraud)
- Develop fully connected policies with approved data/features/frameworks/models
- Centralize approval process and manage decision traceability
- Compare and benchmark policies against baselines
- Execute policies at scale on Spark distributed environment
- Extract production ready artifact for immediate deployment
Cutting Edge Tech Stack
- Corridor is built on Python based open source stack with Spark for distributed computing
- Platform scales easily by augmenting the Spark compute cluster to meet response SLA’s
- Front end UI runs on browser client with Angular / Javascript stack for optimal experience
