{"slug":"inmobi-uses-machine-learning-for-anomaly-detection-in-ad-data-pipelines","url":"https://findausecase.com/use-cases/inmobi-uses-machine-learning-for-anomaly-detection-in-ad-data-pipelines","title":"InMobi uses machine learning for anomaly detection in ad data pipelines","description":"InMobi, a mobile advertising platform, migrated from a multicloud data warehouse to the Databricks Data + AI Platform to unify data warehousing, AI and analytics workloads. Using Spark Declarative Pipelines for anomaly detection, InMobi achieved a 50% improvement in SLAs and an 80% reduction in costs. Teams use MLflow to build their next-generation AI platform and are exploring large language models to let end users query data conversationally. The migration also delivered 32% lower total cost of ownership, 15% faster queries, and 20% better reporting performance versus their prior multicloud data warehouse.","company":"InMobi","industry":"Technology & Software","aiCapabilities":["Fraud & Anomaly Detection"],"technology":["Databricks SQL","Spark Declarative Pipelines","Unity Catalog","MLflow"],"deployment":"Public Cloud","problemStatement":"As InMobi's data processing requirements grew to 20+ terabytes per hour, the cost of running their proprietary multicloud data warehouse skyrocketed, and its closed nature created silos that hindered collaboration and data sharing.","solutionApproach":"InMobi migrated from its multicloud data warehouse to the Databricks Data + AI Platform to unify data warehousing, AI and analytics, using Spark Declarative Pipelines for anomaly detection, MLflow to build its next-generation AI platform, and Unity Catalog for governance, while exploring large language models for conversational data queries.","businessValue":"InMobi saw a 50% improvement in SLAs and an 80% reduction in costs from Spark Declarative Pipelines anomaly detection, plus 34% lower infrastructure costs, 15% faster queries, 20% fewer job failures, a 32% lower TCO and a 24% cost reduction in running ETL pipelines versus their prior multicloud data warehouse.","evidence":{"band":"high"},"sourceUrl":"https://www.databricks.com/customers/inmobi","dates":{"publishedAt":"2026-09-25T05:47:38.037Z","publishedAtSource":"pipeline","updatedAt":"2026-09-25T05:47:38.037Z"},"license":"Open for reading and citing with a link to https://findausecase.com/use-cases/inmobi-uses-machine-learning-for-anomaly-detection-in-ad-data-pipelines. Bulk republication requires permission."}