Mirriad Delivers Next-Generation Ad Tech Using AWS

技术

分析与建模 - 机器学习
基础设施即服务 (IaaS) - 云计算
分析与建模 - 计算机视觉软件

适用行业

汽车

适用功能

销售与市场营销
商业运营

用例

计算机视觉
流程追踪

服务

云规划/设计/实施服务
数据科学服务

挑战

Mirriad, a London-based company, delivers next-generation advertising solutions by using its computer vision technology to naturally place brands in premium video content across TV, online, and mobile channels. The challenge for Mirriad was achieving this intelligent insertion of ads at scale. The company was initially using physical data centers, which proved to be a bottleneck when it came to onboarding big businesses. In addition to fast access to compute capacity, another vital requirement for the company was the ability to use NVIDIA GPUs across multiple regions worldwide.

关于客户

Mirriad is a video technology company based in London that delivers in-video advertising to customers around the world. The company uses machine learning techniques to naturally blend brand advertising into popular entertainment content. Mirriad's technology supports in-video advertising, allowing brands to be placed in premium video content across TV, online, and mobile channels. The company's Chief Technology Officer, Tim Harris, explains that they achieve this by employing machine learning techniques that focus on annotating video metadata so they can pick the right piece of content for the right brand.

解决方案

Mirriad transitioned to Amazon Web Services (AWS), a process that took around a year. This transition instantly solved their capacity challenges and enabled them to set up customers 98% faster. The company uses AWS Auto Scaling across Amazon Elastic Compute Cloud (Amazon EC2) G2 and G3 instances for all of its video processing. This allows them to use between 20 and 40 GPU nodes for a typical production workload, which lasts around 15 to 20 minutes. The company also uses AWS Trusted Advisor to maintain and prove compliance, ensuring that their systems stand up to vigorous security audits.

运营影响

Mirriad can now onboard big clients 98% faster, reducing the onboarding time from up to nine months to just a few days.
The company saves around US$100,000 a year in GPU costs due to the flexibility of AWS Auto Scaling.
Mirriad can now meet the large-scale needs of customers across regions using AWS.

数量效益

98% faster onboarding of clients.
US$100,000 savings a year in GPU costs.
Successful large-scale campaigns resulting in nearly 800 million ad impressions and 58% brand awareness, up 17 percentage points.