Taking over 100 million women towards financial independence through LokOS platform

Streamlining the login process with identity provider implementation using Keycloak:
To address the login issues encountered by admins & other stakeholders, Team Daffodil implemented Keycloak as an identity provider. This transition from database authentication allowed us to delegate user management to Keycloak, which specializes in secure user authentication. By managing user sessions independently, Keycloak reduced the load on the primary application, leading to improved response times and system reliability. Consequently, the platform became more robust and could support a larger number of users without affecting application performance.

Accelerating performance through intelligent cache management using Redis:
Team Daffodil significantly improved application performance by utilizing Redis to cache approximately 120 GB of frequently accessed data in memory, enabling faster retrieval and reducing latency. For Self-Help Group (SHG) admins, this meant that they could complete transactions, view reports, and manage profiles in near real-time minus the frustrating delays.

Efficient queue management for streamlined operations using Azure service bus:
To optimize data handling, Team Daffodil implemented Azure Service Bus, enabling asynchronous processing of user requests. Rather than sending data directly to the database upon user submission, the system now routed transactions through the Azure Service Bus. This method allowed the application to manage user requests more efficiently, minimizing the risk of database overload during peak usage times. Additionally, this reduced the frustrations of waiting for processes to finish, allowing admins to seamlessly continue their tasks.

Enhancing scalability and efficiency with advanced file management:
Azure Blob Storage was used for managing files related to user transactions. This enhanced availability ensured that files were stored reliably and were easily accessible when needed, particularly important for bookkeepers in managing financial records, meeting notes, and other important materials.

Distributed tracing and advanced logging:
The integration of distributed tracing tools significantly improved the ability to monitor the application’s operational flow in real-time. With detailed logs capturing user behavior and system performance metrics, we were able to analyze patterns that might indicate underlying problems. For example, logs enabled us to pinpoint transaction timeouts or errors, helping distinguish between data-related issues and functionality issues. This proactive monitoring ensured that we could address potential problems before they escalated, ultimately reducing user frustrations.

Health check endpoints:
Health check endpoints were established to monitor the performance and availability of services such as Azure Blob Storage, Redis, etc. in real-time. This proactive approach enabled the team to detect and address infrastructure issues before they impacted users.

By accessing real-time feedback on the operational status of these components, Team Daffodil was able to swiftly detect any anomalies and identify any infrastructure issues. This continuous cycle contributed to a more resilient application, ensuring that users come across minimal downtime or disruptions.

Code quality assurance with SonarQube:
The implementation of regular code reviews and quality checks through SonarQube ensured adherence to best coding practices. This resulted in a more maintainable codebase, reducing technical debt and facilitating a leaner application.

Aadhaar and bank account verification:
Aadhaar is a twelve-digit unique identity number that serves as a proof of identity and address for Indian citizens.

The implementation of Aadhaar and bank account verification, along with deduplication processes, was crucial for enhancing the integrity and reliability of the system. With a database comprising of approximately 100 million+ users, ensuring that each Aadhar number and bank account was unique, was essential for accurately identifying individuals within the Self-Help Groups (SHGs).

Team Daffodil digitized the entire manual process of identification, effectively eliminating ghost members and mitigating the risk of fraud and unauthorized access to financial services.

By creating unique IDs against SHG members and offloading the verification tasks to established platforms like DigiLocker and National Payments Corporation of India (NPCI), the LokOS application could maintain its performance while ensuring accurate and secure identity and account verification. This streamlined process not only reduced the burden on internal resources but also expedited the verification timeline, allowing for quicker access to services and support for admins. As a result, the overall efficiency of the system was enhanced, facilitating a smoother user experience while maintaining the integrity and reliability of the data.

Offline capability with regular data synchronization:
The system was designed to function effectively even in offline mode, allowing users to continue their work without interruption. Data collected during offline sessions is synchronized with the server every six hours. This ensures that all information remains up-to-date and consistent, even if the users are temporarily disconnected from the internet. This feature enhances accessibility for bookkeepers in remote areas, ensuring they can reliably input and manage data without being hindered by connectivity issues.

Comprehensive reporting:
State and national systems relied on LokOS platform to retrieve crucial reports. However, the inefficiency of the system led to lengthy processing times, often resulting in failures.

To streamline this process, Team Daffodil employed a data lake architecture to manage the huge amounts of data generated at the time of reporting. Instead of real-time reporting, reports were generated every morning, reflecting data from the previous day. This allowed for efficient data storage and retrieval without directly impacting the operational database every time. By utilizing structured JSON reports based on the data housed in the lake, we maintained system performance and ensured that the operational database remained undisturbed during reporting activities.