Barlow Kasule
← All work

FLT Analytics Platform

Turning scattered transaction data into analytics and regulatory reporting for Uganda's SACCOs and MFIs

Role
Lead Data Science & AI Engineer
When
09/2023 – now
Stack
  • Python
  • FastAPI
  • Pydantic
  • PostgreSQL
  • MariaDB
  • Churn prediction
  • Time-series forecasting

Context

FutureLink Technologies runs SavingsPlus, a banking-as-a-service platform for SACCOs and microfinance institutions, and MSACCO, a consumer app. Both move money through MTN MoMo, Airtel Money, Interswitch and other integrators. Before the platform, reporting, including Bank of Uganda submissions, meant manual Excel work and scattered one-off queries.

Role

Lead Data Science & AI Engineer, and the company's entire data function. I designed, built and maintain the platform end to end.

What I built

  1. Integrators
    • MTN MoMo
    • Airtel Money
    • Interswitch & others
  2. Intake
    • Company portalAll transactions arrive through one route
  3. Data layer
    • FLT serversMariaDB → PostgreSQL (in UAT)
  4. Consumers
    • Internal dashboards
    • Bank of Uganda reports
    • MSACCO in-app reporting
    • Churn & revenue modelsPredictive layer
Every integrator feeds one governed data layer, which serves three consumers and the predictive layer.
  • One source of truth. Transactions from every integrator flow through the company portal onto FLT's servers, where they are consolidated for analysis. The data layer is moving from MariaDB to PostgreSQL, now in user acceptance testing.
  • Three consumers served from the same data. Internal dashboards, automated Bank of Uganda regulatory reports, and in-app reporting for MSACCO users.
  • Predictive layer. Member churn prediction flags SACCO members at risk of going inactive, and revenue forecasting feeds the dashboards.
  • Re-architecture. I rebuilt the original Flask service in FastAPI with a clean, layered architecture (service layer, repository pattern, Pydantic schemas) to modernise it and make room for new features.

I also completed full documentation of the SavingsPlus platform (Laravel, Vue 3) as groundwork for a planned assistant across FLT's products.

Outcome

Manual spreadsheet reporting was replaced by automated pipelines. Regulatory reports now come from the same governed data as the operational dashboards, and the platform scales with a growing subscriber base without adding analysts. All of this sits within FLT's PCI-DSS-certified environment.