SQB Mahalla
A single digital analytics and banking platform built for Uzsanoatqurilishbank (SQB)
Overview
SQB Mahalla is the bank's single digital platform, uniting two jobs: collecting and AI-analysing socio-economic data across Uzbekistan's regions (viloyats, districts and mahallas); and a mobile (PWA) workplace for mahalla bankers that works right next to the client. Through it the bank monitors regional development, delivers lending and business-analytics services to entrepreneurs, and routes applications into the bank's internal systems.
The problem we solve
Traditionally, regional indicators live in scattered sources, and the mahalla banker depends on paper processes when working with a client. The platform solves:
- collecting and visualising indicators for every region in one database;
- on-site client identification (MyID) and electronic loan-application filing;
- generating business analysis and recommendations with AI;
- auto-routing applications into the bank's CRM and internal systems;
- transparent monitoring of bankers' field activity.
Core modules
Analytics dashboard
Shows socio-economic indicators for every region on an interactive map and cards; drill-down from viloyat → district → mahalla.
“Mahalla Banker” mobile app (PWA)
An installable, offline-capable app for the banker in the field: loan calculator, MyID-based application, geotracking, voice recording, and offline save that sends when the network returns.
AI Advisor
An AI module that gives entrepreneurs business ideas and recommendations for a chosen sector and region; answers and a business plan tuned to regional data.
Admin control panel
Manages the reference book of regions, districts and mahallas, indicators and card templates; Excel import; an activity audit log.
Reference book & AI pipeline
Manages the composition and boundaries of regions/districts/mahallas; automatically turns Excel data into indicators.
Reports
Generates reports in PDF, PPTX and DOCX - ready documents by region.
Users and roles
The platform uses a role-based access model; each user acts only within their permissions:
- Administrator - manages the system and users, configures data and keys.
- Editor - uploads and publishes regional data.
- Mahalla banker - on-site identification, loan calculator, application and AI analysis.
- Branch manager - the same functions in the branch (without regional ones).
- Observer - views statistics across the republic/region.
- Client - uploads documents via a personal cabinet.
Data coverage
External integrations
MyID
Biometric client identification.
CRM “Fido Biznes” / IABS
Creates a lead from an application and links to the bank's core system (IABS).
Central Bank (CBU)
Official currency rates.
SMS gateway
Notifications via the bank's internal SMS gateway.
Technology stack
- Backend: Python 3.11, Django 5.2, Django REST Framework (JWT).
- Database: PostgreSQL (primary), Redis (cache & queues).
- Async processing: Celery (background jobs & AI pipeline).
- Frontend: web app + PWA (installable, offline-capable).
- AI: OpenAI (GPT-4o-mini) - analysis, recommendations, chatbot.
- Deployment: Gunicorn + Nginx/Apache; secure HTTPS.
Information security
The platform is protected to the bank's requirements: role-based access control, JWT tokens with session revocation, encryption of secret keys, login-attempt throttling (lockout), an audit log of every important action, and secure HTTPS exchange with external services. A bilingual information-security specification was produced for the project in line with O‘z DSt 1987:2018.
Status and results
- A database covering all 15 regions, 201 districts and 3,234 mahallas is in place.
- The banker platform works: loan calculator, MyID application, AI analysis and CRM integration.
- The AI module produces business analysis and recommendations in real time.
- Geotracking, offline mode and PWA keep the banker working uninterrupted in the field.
Want to know more about this project?
Reach out to the SQB AI team - we'll share the details and show the platform in action.
Contact us