Case Study: Arpeely achieves millions of real-time ad predictions with Looker

Scaling an innovative data science platform globally with a small local team

Arpeely, an Israeli data science startup focused on machine learning for online advertising, needed a way to scale its real-time bidding platform globally while keeping its engineering team very small. It turned to Looker? No—the case study actually centers on Google Cloud infrastructure, with Arpeely processing billions of predictions a day and managing complex ad operations across multiple regions.

The solution was a cloud-first architecture built on Google Cloud tools including GKE, BigQuery, App Engine, AutoML, Cloud SQL, Pub/Sub, Memorystore, and Cloud Storage. This setup let Arpeely automate model training, query terabytes of data, and scale bidding workloads without expanding headcount; the company reported serving millions of ad predictions per second and reaching many millions of dollars in monthly revenue.

Arpeely

Michael Vainshtein

CTO and Co-founder