A leading Fin-Tech lending company in Asia

Challenges

• KYC was a manual process and required the collection of a number of documents. Our team had to study the compliance laws and design an interface that would automate document collection, verification and reduce the total time taken for KYC.
• Underwriting was manual and required the collection of documents for income proof. Our team along with the underwriting officers created machine learning models to calculate income, liabilities and risk.
• Compliance and regulations required complex reports to be automatically generated on a daily, monthly and quarterly basis.
• The system had to be periodically audited by authorities to ensure it complied with laws and met regulatory requirements.

Solution

• Wednesday worked with the customer in a product development capacity. We set up a team of engineers, designers and a project manager to work closely with the customer to build a product that adds value to their business.
• Our design team took an agile approach. Wireframes were built and verified by showing them to users. The team also built a design system that included a suite of components, themes and types. This allowed the developers to build products for mobile and web with the same visual language.
• Our engineering team built the backend using Go. Integrations with payment gateways, third party lenders, account management, loan management, treasury, notifications, reporting and a configuration system to meet the regulatory requirements were some of the features that we built. The backend exposed a RESTful API.
• The frontend for the user facing portal, admin portal and the treasury were built using React.
• The customer facing mobile app was built using the Android SDK. The application was built to be offline first and allowed users to perform actions even when offline. A sync mechanism was built to allow data to sync back to the backend when network was restored.
• The team followed a strict software development process ensuring work was divided into sprints. After each sprint a build was released that was production ready. Each feature had at least 70% test coverage and thorough documentation.

Results

• The use of machine learning models to analyze risk ensured there was constant learning from each loan, which led to a gradual reduction in NPAs.
• Time taken to onboard a new customer was reduced by 80% and completely automated without any manual intervention.
• Card processing and issuance was automated and required no manual intervention.
• Reporting was automated, which led to the reduction in time spent to meet compliance and regulatory requirements.
• Documentation and a configuration system