Unbabel_Customer-Reference_Logo@2x

Unbabel Case Study

2014

Unbabel provides translation as a service through its platform powered by artificial intelligence (AI). By using a mixture of AI and human translation, it can deliver high-quality content five times faster and five times cheaper than its competitors. This is opening up localization opportunities to business of all sizes, allowing them to present content such as frequently asked questions or social media streams in local languages and thus deliver better support and customer service. Unbabel won an Innovation Excellence Award for its translation services at the Translation Automation User Society Conference in 2015.

start a python tutorial
CustomerReferences_QuoteMark

We’re saving money that we can direct to delivering better services, we’re more reliable, and we have a better control of our infrastructure and resources."

João Graça
Chief Technology Officer, Unbabel

The Challenge

To stay competitive in the translation services market, Unbabel needs to have strong technology coupled with solid human resources. “We have our own machine translation engine,” says João Graça, Chief Technology Officer (CTO) at Unbabel, “and a community of more than 40,000 crowdsourced translators, who work in 45 languages and are based around the globe. Combined, they help us produce a fast, cost-effective service.”

But about a year into operations, Unbabel found that its IT infrastructure wasn’t supporting the business adequately. There was frequent downtime, and costs were spiraling, even with a relatively small infrastructure. “We weren’t getting the performance we needed from our cloud platform-as-a-service provider,” says Graça. “Most crucially, the system wasn’t scaling. Plus, it didn’t offer a lot of the services we required for our natural language processing and machine learning technologies. We wanted to work with a cloud provider that could offer us a larger feature set as well as give us greater control over resources.”

Why Amazon Web Services

When comparing its wish list to cloud providers on the market, Graça says it was a “no-brainer” to go with Amazon Web Services (AWS). “We’d already started using AWS for static assets and data backup, as well as some microservices, so when we decided to move our central web application from our main provider, we didn’t think twice about using AWS. It had the depth of services we were looking for, and, after talking to our AWS account manager, we realized the AWS Activate program could give us the support we needed to move wholesale to the AWS cloud,” he says.

After an initial database migration in March 2015 proved successful, the team moved its remaining services to AWS by mid-2015. It uses Amazon Elastic Compute Cloud (Amazon EC2) with Auto Scaling and Elastic Load Balancing to host its main application machines, which sit within the Amazon Virtual Private Cloud (Amazon VPC). It runs production servers in the US East (N. Virginia) region distributed across multiple Availability Zones. It also operates staging and development environments in the EU (Ireland) region. Graça says, “Here, we replicate production environments and test new features. It’s really useful because it gives developers the ability to experiment with different AWS technologies in their region without affecting production environments.”

Amazon Simple Storage Service (Amazon S3) continues to house Unbabel’s static assets and backup data, as well as to store configuration files. It also uses Amazon Relational Database Service (Amazon RDS) for PostgreSQL, and Amazon Route 53 to manage its subdomains.

Amazon CloudWatch allows Unbabel to monitor and set alarms related to worker environments, which run the background jobs in the main Unbabel app. Account activity is tracked through Amazon CloudTrail, and AWS Identity and Access Management (IAM) gives granular control of access to resources, with additional security through two-factor authentication. Amazon Elasticsearch Service forms the back end for Unbabel’s full-text service to customers.

The Benefits

Using Amazon Elastic MapReduce Yelp was able to save $55,000 in upfront hardware costs and get up and running in a matter of days not months. However, most important to Yelp is the opportunity cost. “With AWS, our developers can now do things they couldn’t before,” says Marin. “Our systems team can focus their energies on other challenges.”


About Unbabel

Unbabel provides translation as a service through its platform powered by artificial intelligence (AI).

Benefits of AWS

  • Over $50,000 savings in hardware costs
  • New system up and running in days, not months
  • Increased storage capacity

AWS Services Used

Amazon S3

Amazon Simple Storage Service (Amazon S3) is an object storage service that offers industry-leading scalability, data availability, security, and performance. 

Learn more »

Amazon EC2

Amazon Elastic Compute Cloud (Amazon EC2) is a web service that provides secure, resizable compute capacity in the cloud.

Learn more >>

AWS Auto Scaling

AWS Auto Scaling monitors your applications and automatically adjusts capacity to maintain steady, predictable performance at the lowest possible cost.

Learn more >>

Elastic Load Balancing

Elastic Load Balancing automatically distributes incoming application traffic across multiple targets, such as Amazon EC2 instances, containers, IP addresses, and Lambda functions.

Learn more >>


Get Started

Companies of all sizes across all industries are transforming their businesses every day using AWS. Contact our experts and start your own AWS Cloud journey today.