Case Study: Nanit improves model deployment and production monitoring with Weights & Biases

How Nimrod and the team at Nanit are building smart baby monitor systems, from data collection to model deployment and production monitoring

Nanit, a company that develops smart baby monitoring systems, faced the challenge of deploying and reliably maintaining highly accurate computer vision models on a massive scale. Their models, which track baby sleep patterns and breathing, needed to process tens of millions of requests nightly for over 100,000 users, requiring extreme reliability and a robust process to handle everything from data collection to production monitoring. To manage this complex workflow, they turned to Weights & Biases.

Using Weights & Biases, the team standardized their machine learning process by managing all experiments, comparing model performance, and using reports to facilitate collaboration with their product team. This provided the critical visibility needed to select the best models and deployment strategies. The solution helped Nanit maintain a highly reliable and accurate service, ensuring parents receive trustworthy data and significantly reducing the number of support tickets related to model performance, directly impacting customer satisfaction.

Nanit

Nimrod Shabtay

Senior Computer Vision Algorithm Developer