Case Study: Cognitiv builds better machine learning models with ClickHouse

How Cognitiv Uses ClickHouse to Build Better Machine Learning Models

Cognitiv, a deep learning ad tech company, needed a faster and more efficient way to manage massive data volumes for its offline feature store and machine learning workflows. Its existing data systems were costly, slow to query, and cumbersome to maintain, which limited iteration speed for building and improving bidding models. Cognitiv turned to ClickHouse, using ClickHouse Cloud and an earlier ClickHouse proof of concept to evaluate a better fit for large-scale data processing.

ClickHouse helped Cognitiv run complex scans and joins across huge datasets with much lower latency, turning queries that previously took hours or even days into much faster operations. After migrating roughly two petabytes of data to ClickHouse Cloud, Cognitiv improved efficiency, simplified operations, and reduced the burden of managing its own infrastructure. With ClickHouse, the data team can iterate faster, focus more on model development, and expand ClickHouse use across additional data science and analytics workflows.

Cognitiv

Jason Ruckman

Engineer