Using state-of-the-art human behavior recognition analysis, Realeyes is a platform that measures people’s emotions as they view video content through their computers’ webcams. The firm works with brands, media companies, and market research agencies to deliver business intelligence through emotion analytics. It recruits volunteer participants from around the world to take part in individual studies to test responses to video content. The results of these studies help organizations focus their marketing messages and make content placement decisions that maximize impact. Founded in 2007, with headquarters in London and offices in the U.S. and Hungary, the startup counts among its clients many Fortune 500 companies and some of the biggest brands in the world, including Coca-Cola, Samsung, and Walt Disney.
Initially relying on a hosted services provider, Realeyes received its first round of funding in 2011, at which point the company decided to build its next-generation services entirely in the cloud. Elnar Hajiyev, CTO and cofounder of Realeyes, explains, “We wanted to run everything online. Being a small startup, we didn’t have big resources, and what investment funds we did have we needed to use wisely and put toward adding value to our clients rather than toward IT. Running everything in the cloud made sense.”
Before looking to the cloud, Realeyes was running its services on a single server, which meant any hardware problems could disrupt services and capacity was often an issue. “Operating our internal R&D activities locally meant we’d often use all of the computational capacity of our servers,” says Hajiyev. “This would limit the number of experiments we were able to execute and slow down our research activities.”
The firm needed a scalable infrastructure that could enable research and drive innovation. Security was also key. “We work with sensitive personal data,” says Hajiyev. “Security is the number one consideration for many of our clients.” In addition, global service availability was a critical factor to ensure the firm could run effective worldwide campaigns for its clients near real-time.
Why Amazon Web Services
“The maturity of the Amazon Web Services ecosystem and richness of its cloud services played a significant role in our decision making,” says Hajiyev. And, once the decision was made to move forward, Hajiyev explains that it was easy to get started with AWS. “It took just a few minutes from registering our account with AWS to provisioning our first instance, and less than a week to complete the migration process.” From that point, Realeyes has expanded its use of AWS services from Amazon Elastic Compute Cloud (Amazon EC2) instances to a broad set of technologies from the AWS portfolio today.
The firm makes extensive use of Amazon EC2 for both its webcam video-processing service and scalable machine-learning infrastructure. Realeyes uses both on-demand and Spot instances, and it regularly scales up to hundreds of instances. Amazon Simple Storage Service (Amazon S3) is the main data storage platform, where client video content and terabytes of facial expression data are held.
For efficient content delivery and processing of thousands of concurrent webcam sessions each minute, the firm relies on Amazon CloudFront and EC2 regions. For processing job management and auto-scaling, it employs Amazon Simple Queue Service (Amazon SQS).
Maintenance of SQL Server is conducted through Amazon Relational Database Service (Amazon RDS)—a job which otherwise would have taken a dedicated SQL Server engineer. The company’s media-servers deployment uses Amazon Route 53 for efficient load balancing and proximity based content delivery, as well as AWS Elastic Beanstalk for automated resource management and AWS CloudFormation for managing its latest generation of data collection services. AWS Identity and Access Management (IAM) and Amazon Virtual Private Cloud (Amazon VPC) ensure the company can satisfy strict security requirements.
Realeyes is a global operation whose ability to run major worldwide campaigns lies at the heart of its success. For this, the firm requires the agility that only the cloud can provide. Hajiyev says, “From the start, we gained a tremendous advantage from working in the cloud. The team was spread out across the world, so working in the AWS cloud centralized our development efforts. Using AWS, we were able to build a global data collection service in a few weeks—otherwise it would have taken us months. Being fast to market has helped us secure our position in the field.”
In addition to operating on a global scale, the subject of security is high on Hajiyev’s priority list—as it is for the firm’s clients, many of which are compliant with ISO 27002 and have specific information security needs. He says, “Information governance is a very important topic for us and our clients . With services such as Amazon VPC, Amazon S3 data encryption, and IAM data access authorization, we can clearly demonstrate to them that we’re following the strictest data security principles. This is often the first conversation we have with big clients, and it always puts their minds at ease.”
Being able to accelerate research is key to the firm’s continued success. “With AWS, not only have we reduced the development time for our data collection service from months to weeks, but executing training modules is also several times faster. As a result, we’re quicker to develop new features, which give our clients unique new insights.”
Indeed, AWS has helped Realeyes gain agility in its operations—not only in terms of time to market, but also in the sheer scale it can achieve and the flexibility of pricing. “With AWS, we have built highly scalable services without the need for large upfront investment. For example, our video processing service can automatically scale up to processing millions of participant sessions within hours, plus we’ve saved countless infrastructure development hours and have been able to lower our hardware and system maintenance costs. This is encouraging innovation because we can now plow that money into research.”
Hajiyev concludes with an eye on future projects: “There is much more we can do with AWS. For example, we’re already looking at building APIs purely on AWS Lambda and extending our use of Spark on AWS.” For Realeyes, taking advantage of the full power of cloud computing is key to its ambitions for its emotion analytics technology. “With AWS technologies, we’ve accelerated our research in emotion analytics,” says Hajiyev. “So far, our focus has been on advertising and marketing, but we hope our technology can be used for other applications. We want to continue innovating and bring a human dimension to data-driven decisions across other disciplines.”
“With AWS, we have built highly scalable services without the need for large upfront investment. This is encouraging innovation because we can now plow that money into research.”
Elnar Hajiyev, CTO and Cofounder