Grab updated its internal platform to monitor Apache Kafka data quality in real time. The system uses FlinkSQL and an LLM to detect syntactic and semantic errors. It currently tracks 100+ topics, preventing invalid data from reaching downstream users. This proactive strategy aligns with industry trends to treat data streams as reliable products. By Patrick Farry