MARLO · PRMS · STAR · CLARISA · LLM-driven integrations

The problem

CGIAR runs critical legacy platforms — MARLO, PRMS, STAR, CLARISA — which live in silos. Executive reporting across them was manual, fragile, and slow. Data did not cross-reference; each platform required hand-driven export-import to produce a single combined view.

The approach

I led the intelligent integration between these platforms and a layer of AI services. The pattern is clear: APIs and webhooks as the bridge, custom agents consuming and producing events, and a pipeline combining LLMs and RAG to extract and harmonize data without asking the teams to change the systems they live in.

• Webhooks and REST connectors to receive events from each platform.
• Agents that query, transform, and publish data cross-system.
• Standardized delivery on Agile/Scrum, CI/CD on Jenkins, DevOps practices, and design-first workflows (Figma + Business Analyst).
• Specialized QA and user-support roles to close the operational loop.

The outcome

• Dramatic reduction in executive-reporting cycle times.
• Cross-system intelligence where there used to be manual export-import.
• Stronger operational excellence and user experience thanks to the new QA and support roles.
• I mentored the team and championed sustainable delivery: balanced workload, better project management, better retention.