Building the engineering foundation for CrediPay.

Credix Finance’s CrediPay product brings credit into business transactions through an API. It allows sellers to offer payment terms to their buyers, with credit analysis built into the process. Supporting that product meant handling invoice submissions reliably while coordinating the checks needed to secure credit.

Robonaut joined while Credix was still an early-stage company. Its existing platform had grown from a blockchain proof of concept, and the application and operational foundations needed to develop alongside CrediPay’s ambitions. The work connected technical direction with hands-on implementation across the software, infrastructure and the way the team delivered changes.

Infrastructure automation was an early priority, but it did not resolve the whole delivery problem. Work could still spend months on long-running branches before reaching production, making releases stressful. The goal was to give the team the confidence and means to release daily.

We built a delivery foundation around automated versioning, immutable artifacts and recorded promotion between environments. Terraform brought infrastructure into code, while moving from the existing App Engine setup to containers on Cloud Run gave the Python risk services and Node.js API a common approach to builds and deployment.

The engineering practice changed with the tooling. Work was considered complete when it had been deployed and tested in a running environment. This connected implementation to its behaviour in use and made delivery part of everyday development.

From a proof of concept started in late 2024, CrediPay reached its 100th production release by August 2025. Daily releases became an established practice that the team continues to follow.

To understand where the platform needed to change, we introduced searchable logs and application performance monitoring across the Python, Node.js and PostgreSQL stack. This exposed expensive database queries and gave the team a measurable basis for optimisation. Customer reports and operational data increasingly informed the technical priorities.

To prepare for the business’s ambitious growth plans, Credix set an internal capacity target of more than 100 invoice capture requests per second. Analysis showed that third-party services and slower credit checks constrained the synchronous request flow. We built an event-driven v2 API that separated invoice validation and preparation from the backend processing. The API could accept work without waiting for every downstream check to finish.

That separation allowed request handling and credit processing to progress independently. Load testing in a production-like environment achieved the target of more than 100 invoice capture requests per second, with traffic analysis and performance monitoring showing how the system behaved under load.

The new API changed the integration contract, so the transition was staged around customers. Existing integrations continued on the synchronous v1 API while v2 was tested through automated end-to-end checks and load tests, with logging and monitoring already in place. Smaller new customers adopted v2 first, followed by existing customers as they updated their integrations.

This gave the team a gradual path away from v1 while keeping the existing service available during adoption. Shared delivery practices across the API and risk services also made the platform more consistent for the engineers building and operating it.

The contribution brought architecture, implementation and operational practice together. Credix had a tested path to higher volumes, repeatable releases and the visibility to guide further improvements as CrediPay developed.