Loan Approval Workflow
Faster Decisions for CreditNinja
- Client
- CreditNinja
- Industry
- Consumer Lending / FinTech
- Service
- Third-Party Integration Engineering | AWS Lambda + Chalice | Python Multithreaded Decisioning | CI-Ready Test Engineering
Overview
CreditNinja is a fast-growing web lending and approval application. It uses a loan-approval workflow to facilitate transactions for all parties.
A number of third-party services needed to be efficiently integrated in order to fetch and store the data required for the credit-decisioning process — without slowing the loan-approval workflow that CreditNinja’s customers depend on.
How We Solved It
Sphere built the integration layer as a fleet of Python 3.6 AWS Lambda functions, deployed with the Chalice framework. The decisioning service used Python’s built-in multithreading to call those functions in parallel, so data from third parties was available faster than a sequential pipeline could deliver. The codebase was kept lightweight by leaning on built-in packages, and shipped CI-ready with 95–100% test coverage.
1. AWS Lambda Functions in Python
Each third-party integration was isolated as its own Lambda function in Python 3.6. The model lets each integration scale and fail independently — a slow vendor never holds up the rest of the workflow.
2. Chalice for Versioned, Zero-Time Deployments
The Chalice framework gave us versioned, zero-time deployments and clean dependency packaging — turning operational risk into a non-event when functions need to roll forward or roll back.
3. Multithreaded Decisioning Service
A central decisioning service calls the integration functions in parallel using Python’s built-in multithreading package. Data arrives faster, decisions go out sooner, and the customer waits less.
Key Outcomes
95–100% Test Coverage
CI-ready code with 95–100% test coverage made the integration layer materially more reliable than what it replaced.
Improved Decision Model
Sphere’s integrations let CreditNinja improve its decision model — untrusted customers are filtered out earlier in the application flow.
Significant Financial Savings
Catching higher-risk applications early translated into real financial savings and a meaningful drop in default-driven losses.
Faster, Lighter Deployments
Built-in package usage kept the codebase lightweight, so deployments are quick and on-call maintenance stays painless.
The Results
Sphere’s integrations let CreditNinja improve their decision model — significant financial savings and risk reduction, because untrusted customers are filtered out early in the application stages. The reliability of the system also went up materially, thanks to disciplined test coverage and a deployment story built for speed and safety.
