TCS · GNN · agents · 2026

Train-delay forecasting.

A proof of concept that forecasts train delays from live operational data with a Graph Neural Network.

Graph Neural Networks
AI agents
SQL triggers
Event-driven pipelines

Problem

An Indian Railways (IRCTC) client pilot needed delay forecasts from operational data as it arrives, not from a nightly batch.

Build

An event-driven pipeline: SQL triggers on the client's operational database flag new records, an AI agent picks them up and converts them into graph-structured inputs, and a Graph Neural Network inference engine forecasts the delays.

Result

Delay forecasts from live operational data (proof of concept, in progress).