Case Study: Senseye achieves 120x faster machine learning with Saturn Cloud

Senseye uses Saturn Cloud to train machine learning models on GPUs at a massive scale

Senseye, a company specializing in neuroscience and computer vision, faced a significant challenge with the immense computing power required to process high-definition video data for training its machine learning models. Using its on-premise infrastructure, one compute-intensive processing pipeline took 60 days to complete, severely limiting its ability to iterate on solutions and improve its product. This bottleneck forced them to make decisions with incomplete information.

To overcome this, Senseye partnered with Saturn Cloud, leveraging its data science platform to scale their processing on AWS. By using Saturn Cloud to scale to 160 T4 GPUs in the cloud with only 10 lines of code, they drastically reduced processing time. The solution from Saturn Cloud enabled them to process videos 120 times faster, slashing the total runtime from 60 days down to just 11 hours. This performance improvement allowed for many more iterative model tests, significantly enhancing the effectiveness and robustness of their AI-driven product.

Senseye

Seth Weisberg

Principal Machine Learning Scientist