We needed a way to detect ingestion anomalies across 200+ microservices without building a separate ML stack. It turns out BigQuery already tracks every row written in WRITE_API_TIMELINE, and its built-in AI.DETECT_ANOMALIES (powered by Google’s TimesFM) can handle the detection.
We needed a way to detect ingestion anomalies across 200+ microservices without building a separate ML stack. It turns out BigQuery already tracks every row written in WRITE_API_TIMELINE, and its built-in AI.DETECT_ANOMALIES (powered by Google’s TimesFM) can handle the detection.