An owned digital product in a market with fragile infrastructure.

We designed and built a credit platform for financing vehicles, motorcycles, trucks, and agricultural machinery in a market with limited banking infrastructure, high informality, and complex currency dynamics.

The challenge

Consumer credit in this market doesn't fail for lack of demand — it fails for lack of infrastructure: traditional credit scoring is non-existent for much of the public, banks are barely integrable, and the customer decides on the monthly payment, not the rate. A generic e-commerce platform solves none of this.

Architecture applied

We didn't adapt an existing product: we built one of our own on a sober, maintainable base.

• Backend on Django + Django REST Framework, PostgreSQL, and JWT auth — a core a small team can operate without depending on a vendor.
• Two financing modalities modeled as business rules, not code branches: programmed purchase with award and award credit, including a plan specific to the agricultural segment.
• Financing calculator with an amortization table and a saveable simulation: the customer understands the commitment before requesting it.
• An on-time-payment points system as behavioral alternative scoring, given the absence of a traditional bureau.
• Admin panel for payment validation, plan configuration, and request tracking with visual states.
• Deployed on owned infrastructure (Nginx, Gunicorn, SSL) and direct contact via WhatsApp, where the customer is.

Model layers activated

Data capture · CRM with golden record · Contact center and digital commerce. The platform is, in itself, the model's first layers applied to an owned financial product.

Results

It's an owned product, active. The key architectural decision —behavioral scoring instead of waiting for a bureau infrastructure that never arrives— is what makes it viable to extend credit in this market without pricing the risk away.

Lessons

In fragile markets the constraint isn't the frontend: it's the risk model. Building the calculator and the request flow was the fast part; designing scoring that works without the data a mature market takes for granted was the real architectural work.