Case Study: Sprinklr reduces machine learning inference costs and improves latency with Amazon Web Services

Sprinklr Reduces Machine Learning Inference Costs on AWS Inferentia

Sprinklr, which runs its Unified-CXM platform on machine learning models that process unstructured data from more than 30 channels, needed to improve inference performance and lower costs as it handled about 10 billion predictions a day across 500+ models. The company used Amazon Web Services, specifically Amazon EC2 Inf1 Instances powered by AWS Inferentia, to support its latency-optimized and throughput-optimized workloads.

Amazon Web Services helped Sprinklr benchmark and migrate models to AWS Inferentia, reducing latency by more than 30% on latency-focused workloads and enabling the company to move all latency-optimized workloads to Inf1. After migrating about 20 models, Sprinklr expanded to computer vision and text models, and can now deploy a model on Amazon EC2 Inf1 Instances in under two weeks, improving efficiency, customer satisfaction, and cost savings.

Sprinklr

Jamal Mazhar

Head of Infrastructure and DevOps