Case Study: a construction-tech company reduces labeling and inference costs with Sphere Partners

construction-tech company cuts SageMaker inference costs by 40% with Sphere Partners

Sphere Partners worked with a construction-tech company that was not achieving its desired cost-basis after recently starting with Amazon SageMaker on AWS. The customer was also looking to replace its segmentation model in an effort to reduce its infrastructure costs.

Sphere Partners conducted a comprehensive checkup and optimization of the SageMaker instance. The solution involved updating endpoint configurations, building a SageMaker Ground Truth pipeline to tag images, and implementing AI-assisted labeling. This reduced SageMaker inference costs by 40%, cut the cost of labeling by 70%, and replaced the segmentation model with a more efficient one that reduced output size by a factor of 100.