Case study

Meltwater

Meltwater was the first company to introduce digital media monitoring, and is now the first to apply AI-driven media intelligence.

About

Meltwater is a Global SaaS company that helps companies make better, more informed decisions based on insights from the outside. More than 25,000 companies use the Meltwater platform to stay on top of billions of online conversations, extract relevant insights, and use them to stay ahead of their competition.

Meltwater

Meltwater
Media monitoring

FOUNDERS: Jørn Lyseggen
LOCATION: San Francisco, CA
TIMELINE: 2016 and still ongoing
SERVICES: Backend, Frontend, Infrastructure
OUR TEAM: 3
WEBSITE: https://www.meltwater.com/

Our Mission

Meltwater asked us to rebuild their Facebook monitoring tool, Likealyzer, which required a new UI and backend. The tool attracted a significant amount of users and we soon realized that we had to go beyond these changes in order to keep up with the traffic. That was the start of an exciting journey, leading to the implementation of new algorithms and a state-of-the-art infrastructure built from scratch.

A coherent analysis and evaluation of existing algorithms led to the redesign of the architecture and infrastructure, providing additional features and overcoming big data challenges.
Working hand in hand with Meltwater turned ideas effectively into valuable features and ended in a faster, more scalable and reliable software.

Main challenges

Data

With over one million Facebook pages analyzed, the tool required a solid infrastructure, allowing a smooth performance despite high traffic.

Zero downtime

High website traffic around the clock demands continuous availability of all features. Given the short release cycles, we designed a deployment strategy that guarantees zero downtime when rolling out new features.

Development

The backend was implemented in different modules, all built in Node.js. We worked with Express.js to provide a RESTful API running on AWS Elastic Beanstalk with Docker, built several functions on AWS Lambda and stored data in DynamoDB and RDS.

The frontend was built using React and Next.js framework for server-side rendering, delivering content with Amazon CloudFront.

Apart from those already mentioned, we also worked with other services of the Amazon stack such as API Gateway, S3, Route53, AWS Certificate Manager and CloudWatch.

The use of data science and Machine Learning in Likealyzer is twofold. First, we applied ML to find patterns in the interactions of the users with the tool, in order to discover in which features we should focus the most. Second, we are storing all relevant data that will be used as input for ML models in the future, so we can provide better insights and user experience.

As a result, we were able to build and launch a scalable, secure and highly available product. Now Likealyzer has analyzed over a million Facebook pages and counts with 50k registered users.

Read more

Testimonials

More of our work

Take a look at some examples of our most recent work.

Next case

MercadoLibre

arrow

Ecommerce & auction platforms. Machine Learning & UI.

Learn more

O3b Networks

arrow

Global satellite service provider. Backend & Business intelligence.

Learn more

Lively

arrow

Connected home health platform. Frontend, Backend & Data analysis.

Learn more
Need to boost your product?
We are ready to help you.
Contact us

Get in touch

What project are you currently working on?
Tell us a little bit about it.