Self-Serve Platform for Lenders to Price Products

The Company

Retail Banking Client, USA

The Project

The client's customers are some of the major financial institutions around the world. These customers have similar goals but different requirements on how they want to achieve these goals. Some only want to find optimal price for Mortgages, while others want to use the platform for Auto Loans. The client wants to create a flexible software that could serve the customer's requirements from a single self-serve platform. This way the client's customer of any size can utilize the platform and tailor it to their needs.

Tech Stack
  • NodeJS

  • React

  • Material-UI

  • MongoDB

  • Python

  • Docker

  • Google Cloud

The Goals

While working with Senior Product Executives, below were some of the key requirements that defined the acceptance criteria

  • Every financial institution has a very different schema of their data and we wanted to embrace that difference so that more customers could use the platform.

  • The customers may want to update the data schema to experiment new pricing models without affecting the version of data and associated financial calculations that are already in effect at the bank branches.

  • The customers wanted the ability to try different pricing models with different data schema versions so that they can find a winner combination for a data schema version and pricing model

The Approach
  • The initiative started as a hands-on working sessions with Product Owners to implement a single customer use case using spreadsheets.

  • Once defined, we worked on finding different application designs which can meet the customer needs.

  • Once an approach was defined, the team worked with Senior Data Scientists to create a Pricing Model that could produce the same results as the spreadsheet solution.

  • Once this task was done, we designed flexible data model schema so that every customer could own their own data schema and multiple versions of their schema

  • Then, the application logic was created to support the business operations, use-cases and validations to enable CRUD operations with the flexible data model.

  • The API driven approach helped to separate the UI/UX work with the backend development.

  • Define and document the outcome of the project by working with multiple stakeholders.

  • Design and develop the flexible data model.

  • Architect the solution to replicate the workflow defined by the spreadsheet solution. This was designed to keep in mind the scaling challenges, the transaction boundaries, data privacy, deployment model, and maximize the use of expensive resources such as CPU.

  • Worked with Data Scientists to provide clear API specification of the data needed to perform machine learning computations.

  • Designed the implemented the UI/UX for the workflow that client's customer would use.

  • Created training videos and documentation to enable team members joining the team.

The Impact
  • The solution proved that the client can securely maintain the customer's data schema versions in database instead of having just one version living in source control system

  • The solution demonstrated that the flexible data model could serve multiple use cases for a number of customers.

  • The solution proved that the customers could leverage self-serve platform to do most of their work by themselves than hiring resources to perform manual requests by the customers.

  • The solution proved that the expensive computations could be delegated to serverless solutions and pay per use instead of allocating dedicated expensive cloud instances which may not be fully utilized.

The Solution
  • The solution was deployed as a software platform.

  • It enabled every customer to own their data schema. They could also version their schema and publish new versions as needed.

  • The customers could experiment their ideas using the platform without committing to a specific schema version or a pricing model.

  • The customers could run multiple pricing models against a specific data schema version.

Team Structure
  • 2 Product Managers

  • 1 VP, Product

  • 1 VP, Engineering

  • 1 Data Scientist

  • 2 Developers

  • 2 QA