MARLO · PRMS · STAR · CLARISA — platform architecture & delivery

The problem

CGIAR runs a family of research-reporting platforms — MARLO, PRMS, STAR, CLARISA — that had to be built to one engineering standard, integrated with each other, and deployed reliably across global environments (production, testing, on-prem campus).

The approach

I led the platforms as architect and set the rules of the game:

• Engineering standards: unit testing, static analysis with SonarCloud, code review, clear development practices.
• CI/CD on Jenkins, deploying to AWS (with Azure for some dashboards and an on-prem environment at the CIAT campus for testing).
• A Node.js + Angular + MySQL stack across the platforms.
• Agile/Scrum on Jira — dailies, sprint planning, sprint review, retrospectives — run as a real team cadence.

Architecture

The production environment spans three data centers — AWS (Virginia, global production), Azure (dashboards and data-out), and on-prem (Palmira, CIAT campus, for testing) — with structured DATA-IN / DATA-OUT flows between the platforms and their consumers. (The hero shows the full topology.)

The outcome

• Multiple research platforms delivered and operated to one standard.
• Reliable CI/CD across global production and on-prem test environments.
• Cross-system integrations that replaced manual export/import between platforms.

Reference: the MARLO architecture.