CSMS
CSMS handles container and vessel operations for KSA. I worked on both sides of the system — the operations backend and the screens the operations team uses every day — and built the AI layer that watches shipment data for anomalies.
- Role
- AI Engineer, Fullstack Developer
- Discipline
- Logistics, AI
Built with
- Next.js
- PHP / Laravel
- Python
- PostgreSQL
- REST API
The company already had policies covering many of the situations the system needed to flag. Replacing those with a model would have meant re-deriving knowledge that was already written down, and giving up the ability to explain a decision.
So the AI layer is hybrid. Business rules cover the cases with existing policy. A trained model handles the patterns those rules miss — mainly anomaly detection across shipment data. When something is flagged, it is always clear which of the two did the flagging.
Operations backend
Container and vessel operations, from data model through to the REST APIs the frontend consumes.
The screens themselves
The interface the operations team works in daily, built in Next.js against those APIs.
Anomaly detection, served internally
The model is exposed through internal APIs so predictions appear inside the existing workflow rather than in a separate tool nobody remembers to open.
Next project
IMS
