With Amazon Machine Learning, we provide users and customers with financial intelligence at scale. As a result, our work helps level the playing field, making everyone’s ad revenue, spend, app financials, and job salaries available. 
Dr. Iddo Drori Founder and CEO

AdiMap is a data science company that combines the disciplines of computer science, statistics, and business. The company develops algorithms and software to infer financials, building the world’s first financial graph connecting people profiles, companies, jobs, schools, locations, apps, advertisers, and publishers. The company’s products include AdiMap Jobs for employees and employers, which computes the salaries of open and current jobs in U.S.-based companies; AdiMap Apps for business, which estimates the financials of apps and developers worldwide; AdiMap Spend for advertisers and publishers, which calculates ad spend and revenue worldwide; and AdiMap Elections for voters and candidates, which computes the ad spend of U.S. presidential candidates.  

When AdiMap was founded in 2011, it needed to cost-effectively meet significant compute needs. “We were faced with the challenge of processing a click-stream with very high throughput,” says Dr. Iddo Drori, CEO and founder of AdiMap. “That was going to require massive infrastructure, an entire team of system engineers, and millions of dollars—all of which we didn’t have as a startup.”

Recently, AdiMap looked for a robust technology to increase its machine learning capabilities, particularly for a large enterprise customer. “In online advertising, each network knows its own publishers’ revenue and advertisers’ spend, and we needed a solution that covered all networks,” Dr. Drori says.

The machine learning solution also needed to be highly scalable. “To provide anyone with access to everyone’s ad spend data, we needed to extrapolate from a small sample used for measurement purposes to cover all advertisers and publishers,” says Dr. Drori.  

Adimap decided that the cloud was the best way to meet its needs for cost savings, scalability, and speed to market, and chose to use Amazon Web Services (AWS). “In 2011, AWS was the only viable cloud solution for us, and the early beta AWS services fit our needs perfectly,” says Dr. Drori. One of those services was AWS Elastic Beanstalk, which the company uses to deploy and scale web applications. “We used AWS Elastic Beanstalk, which was then in early beta, to process data at scale for only a few thousand dollars a month. We were able to do that with a team of just three people without investing in new resources or hiring a teams of system engineers.”

AdiMap then expanded its use of AWS to include Amazon Kinesis, a platform for real-time processing of streaming data. The company set up a service that crawls the web, using Kinesis to ingest online ad data and job feeds. The data is processed and sent to Amazon Redshift, a petabyte-scale data warehouse. “We were very early adopters of Redshift and then Kinesis,” says Dr. Drori. “Our Redshift database provided our company and our customers with a high-capacity storage solution at an affordable cost.”

Most recently, AdiMap began using Amazon Machine Learning, which helps companies easily train and use machine learning models for prediction and inference. Using Amazon Machine Learning, AdiMap built predictive models that extrapolate financial ad data for all publishers and advertisers, on all ad networks. As a result, AdiMap was able to provide data measurement services to one of its customers, a large Internet company. The prediction is based on a small random sample of users’ navigational trails, which is extrapolated to the entire population. AdiMap uses a similar framework for computing the advertising spend of presidential candidates, and it solves a large system of equations for computing the salaries of open and current jobs in U.S. companies.

AdiMap plans to migrate its service from AWS Elastic Beanstalk to AWS Lambda, a compute service that enables companies to run code as required, without having to provision or manage servers.   

AdiMap’s goal is to be able to meet its compute requirements without spending significant amounts on hardware resources. “Once we migrate to AWS Lambda, we project that we will significantly reduce costs, which will have a large impact on our business because it will free our resources to scale our data offering,” says Dr. Drori.

The company is using Amazon Machine Learning to gain the scalability it needs. “With Amazon Machine Learning, we provide users and customers with financial intelligence at scale,” says Dr. Drori. “As a result, our work helps level the playing field, making everyone’s ad revenue, spend, app financials, and job salaries available. This gives users and customers a competitive advantage.”

AdiMap has reduced its time to market by using AWS to support its financial intelligence platform. “With Amazon Machine Learning, it’s easy to set up models for prediction and inference instantly,” says Dr. Drori. “Amazon Redshift also helped us speed the development and rollout of our platform to our customers. We opened up the database to our clients and were able to provide direct access. Perhaps most importantly, we were able to keep our team in the loop at each stage of development, and keep our customer updated with the latest results right away.”

In the future, AdiMap plans to launch new products faster by using additional AWS services such as Amazon QuickSight, a business intelligence service used to build visualizations and perform ad-hoc analysis. “We will be using Amazon QuickSight as a front-end visualization tool for Redshift, which is going to give us a faster, easier way to create and test new user experiences and interfaces,” Dr. Drori says. “It usually takes about a month for us to build a working prototype, from idea to design, front end and back end. However, with QuickSight, we hope to be able to reduce that time from weeks to hours. We’re very excited to lower our time-to-market for some of our upcoming products. New AWS services are helping enable our business.”

To learn more about how AWS can help you manage your big data environment, visit our Big Data details page.