Case Study: Humen.Ai achieves 70% infrastructure cost savings and faster releases with Rackspace Technology (Onica)

Humen.Ai dances its way to 70% savings

Humen.Ai, maker of the iOS app Sway: Magic Dance, uses deep learning to generate photorealistic dance videos from short user clips. Its backend relied on hundreds of expensive AWS G4 on‑demand instances and a complex, hand‑built system, which became costly to run, hard to scale and slowed releases—threatening the startup’s financial stability and agility.

Onica containerized Humen.Ai’s AI pipeline onto Amazon ECS with Spot Instances and serverless components, adding a custom ECS scheduling approach to increase machine utilization. Delivered in six weeks, the new architecture cut potential infrastructure costs by about 70%, doubled–tripled processing throughput, reduced technical debt and engineering overhead, and enabled faster releases and reinvestment in product innovation.

Humen.Ai

Tinghui Zhou

Co-founder and CEO