Case Study: mpathic builds faster, leaner ML workflows with ClickHouse Cloud

How mpathic built better ML workflows by switching from Elasticsearch to ClickHouse Cloud

mpathic, a 20-person pre-Series A startup, uses AI to analyze therapy session audio for patient safety and clinical-trial compliance. Before working with ClickHouse, the team relied on Elasticsearch and EC2-based pipelines, which made joins, aggregations, and experimentation slow, costly, and frustrating for its machine learning workflows.

By switching to ClickHouse Cloud, mpathic turned ClickHouse into the backbone of its ML infrastructure, running string extraction, label aggregation, undersampling, and other pipeline steps directly in the database. The result was faster, leaner development: pipelines dropped from about 15 minutes to 4, EC2 usage was eliminated, and engineers spent far less time on infrastructure and more time improving models and analysis.

mpathic

Caraline Bruzinski

Senior Machine Learning Engineer