Automotus - Data Warehouse & BI Implementation

Challenge

Automotus, a computer vision company for curbside management, was processing millions of parking events daily but couldn’t analyze the data effectively. Their PostgreSQL database was hitting performance limits, queries were taking hours to run, and they couldn’t provide real-time insights to city partners about parking violations and curbside usage patterns.

Solution

We migrated their data infrastructure to Snowflake, implementing a multi-cluster warehouse architecture to handle both real-time ingestion and analytical workloads. Using Matillion and Fivetran, we built robust ELT pipelines to process computer vision data, parking events, and violation records. We implemented DBT for data transformations and created a semantic layer for consistent metrics. Finally, we deployed real-time dashboards in Tableau that city officials could use to monitor curbside activity and enforcement metrics.

Results

• Reduced query times from hours to seconds
• Enabled real-time curbside analytics for 5 major cities
• Processed 10M+ parking events daily without performance degradation
• Improved violation detection accuracy by 45%
• Saved $200K annually in infrastructure costs