Rural women's self-help groups record savings and loans, even offline, and reach bank credit with verified records.
The Ministry of Rural Development wanted every member of a rural self-help group to have an accurate record of savings, loans and repayments, so women could reach financial help with ease.
What changed
What stood in the way.
- The platform held millions of records for thousands of groups, and a monolithic design tied the app to one overloaded database that slowed down at peak times.
- Logins ran through the database itself, which was a security risk and made onboarding and access hard to manage.
- Weak verification let duplicate or wrong entries in, so reports needed manual checks before states and the centre could trust them.
- Banks and government programmes depended on timely records for eligibility and loan processing, so any delay hurt them.
- 90%better performance, at most (as reported)
- 100+ millionhouseholds served (as reported)
- 10M+self-help groups managed (as reported)
- 2 trillionin transactions digitised each year within self-help group networks (as reported)
What we built.
- An audit of the code, then the application separated from the database and the database restructured for faster retrieval.
- Logins moved to Keycloak, and around 120 GB of frequently used data held in Redis.
- Requests processed through Azure Service Bus queues, files kept in Azure Blob Storage, on Kubernetes that scales with the load, cutting database load by nearly one-third.
- Aadhaar and bank account checks with deduplication through DigiLocker and NPCI, and a unique ID for every member.
- Offline use for bookkeepers in remote areas, syncing with the server every six hours.
- Daily reports built from a data lake, so the working database is never slowed by reporting.
- Public APIs for state missions, fintechs and partner platforms, plus tracing, health checks and SonarQube reviews.
- Kubernetes
- Keycloak
- Redis
- Azure Service Bus
- Azure Blob Storage
- DigiLocker
- NPCI
- SonarQube
What this work shows about us.
Savings, loans and repayments of self-help groups are the records banks rely on, so we made them accurate and unique at the source.
One platform serves over 100+ million households, and still works offline for a bookkeeper with no signal.
Logins, caching, queues and reporting were each moved off the overloaded database, which is what took its load down by nearly one-third.
Want to rebuild a financial inclusion platform at this scale? Tell us what you want to change. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with the right team.
Start your plan ↗If we built it today, with AI.
This work came before today's AI. If LokOS (Ministry of Rural Development, Government of India) asked us now, this is what we would add, step by step.
An agent would tell a bookkeeper at once which member details or bank documents are missing or do not match, before the record is saved. The Aadhaar and bank checks still run as code.
An agent would put together a group's savings and repayment history for a bank loan application and show the group what the bank will ask. The bank still decides on the loan.
An agent would notice when a group's repayments start to slip and alert the cluster federation staff early. People decide how to help the group.
Where this work fits.
- LendersYour customers are meant to get the loan they need, without chasing anyone.
- Digital lending & loan managementBorrowers get the loan they qualify for without chasing: what's missing and the decision the same day, and the money once it's cleared.
- POS Shop Ltd.Shoppers order from their local store on their phone and redeem loyalty points and store credit in the same app.
Facts and figures from the project records; nothing added.
Want this for your business?
Tell us what you run or what you're building. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with a senior AI engineer.
Send us a note instead
An AI agent. This prototype uses prepared answers and sends nothing.
We use essential cookies to run this site. With your consent we also use analytics cookies to see which pages help. Privacy policy