Case Study: Zencity achieves 20X more experiments at 30% lower cloud costs with MissingLink

How Zencity Accelerated Training from 5 to 100 Experiments Per Project

Zencity, a city planning technology company, uses deep learning to help cities better understand citizen sentiment across social media and other sources. As its data science team grew, it struggled with a largely manual workflow for running experiments, managing configurations and artifacts, and babysitting jobs on cloud GPUs, which consumed about 30% of the team’s time and slowed training at scale.

To solve this, Zencity implemented MissingLink’s AI infrastructure and experiment management platform on Microsoft Azure. MissingLink automated DevOps tasks, enabled queue-based and parallel experiment execution, and kept data securely within Zencity’s cloud account. The result was a jump from 5 to 100 experiments per project, 30% of data science time freed up from DevOps work, and 30% lower cloud costs when running experiments.

Zencity

Ido Ivry

Chief Technology Officer