Building a centralized data platform for a commodity trading firm's China business

Plant data existed, but nobody could act on it in time

The client's China commodity trading business ran dozens of plant processes generating steady streams of operational data, but that data wasn't captured or reported in a way business users could use. Traceability reporting for finished goods, in particular, was a heavily manual process prone to delay and error.

• No centralized platform to capture and report on plant-level operational data.
• Traceability reporting for finished goods was manual, slow, and error-prone.
• Business users lacked visibility into workflows and events happening across the business day-to-day.

Model the upstream process, then build the platform around it

I defined the product strategy, roadmap, and execution for a central data platform, starting by understanding the upstream plant processes in detail before designing how data should flow through the system.

• Understood upstream processes and performed domain modelling to guide data engineers in building the right pipelines.
• Captured 70+ distinct workflows and over 1,000 events per day into the platform.
• Automated traceability reporting for finished goods, removing the manual reconciliation work business users previously did by hand.
• Managed a team of data engineers and partnered closely with data analysts to support downstream BI use cases.

Faster, more reliable data — and a lot less manual work

The platform gave the business a single, trustworthy source for plant and traceability data, freeing business users from manual reconciliation and giving analysts a foundation for further BI work. The team's Gen AI exploration on top of this platform work was later recognized as a hackathon semi-finalist.

Selected views

A look at the workflow capture pipeline and the traceability reporting dashboard.