Sandor Caetano

Chief Data Scientist, iFood

Supernova Award Category: 

AI and Augmented Humanity

The Organization: 

Founded in 2011, iFood, the leading foodtech in Latin America, maintains operations in Brazil, Mexico and Colombia, and in its early years, improved its operating model, following the evolution of the market and growing year by year.

The company receives 30.6 million monthly orders. It works with partners with initiatives that bring together business intelligence and management solutions for the approximately 160,000 restaurants registered in more than 1,000 cities across Brazil. iFood has important investors, such as Movile - a global leader in mobile marketplaces - and Just Eat - one of the largest online ordering companies in the world.

The Problem: 

The company experienced a meteoric evolution (tripling annual growth in 2015-2016), which brought the need to move to the cloud. In preparation for the migration, the IT team rewrote their monolithic systems for microservice architecture. In 2016 they moved all-in to AWS and achieved the flexibility and elasticity they needed. During peak hour operations (lunch and dinner on Sundays) iFood can exceed 500 instances, managing 50 requests per second.

Initially, iFood worked as an integrated network for placing online orders from the restaurants that comprised it. Over time, the company started to develop its own technology. Today, more than 20% of orders placed use the entire iFood sales platform, including the marketplace, CRM and own delivery person.

Just over a year ago, iFood decided to create its Artificial Intelligence Academy, focusing on the development of research in the areas of machine learning, deep learning, logistics efficiency and others related to the company's ecosystem. The initiative is the result of an investment of US $ 20 million, part of an investment made by shareholders in 2019.

In order to carry out the developments, carry out the necessary tests and put them into operation, iFood needed a robust IT structure, which, however, would only be fully used during peak times. The option was to look for a company that offered this infrastructure in the cloud, enabling its payment for use and offering flexibility and scalability when necessary.

The Solution: 

iFood already used AWS services, such as storage and managed database, among others.

Among the main solutions used are Amazon SageMaker, which houses the entire model testing infrastructure; Amazon Kinesis, which facilitates the collection, processing and analysis of streaming data in real time; and Amazon Elastic Kubernetes Service (Amazon EKS) to keep models in the air.

The use of AWS infrastructure allowed the iFood Academy of Artificial Intelligence to expand the use of algorithms. One of the first areas benefited was Logistics. The simulations are performed in the secure AWS environment and, once proven, are put into production.

The iFood app recommends restaurants and dishes according to the user's taste through models processed on AWS. The company is able to control what will be shown in the app, strategies and promotions thanks to these algorithms.

The Results: 

iFood has been achieving significant gains in productivity and improvements in its service levels. In the logistics area, the distance traveled by delivery personnel has been reduced by 12% thanks to the optimization of routes. The same happened with the downtime of these professionals, which was reduced by 50%.

Metrics: 

The company had gains such as: The company reached more than a thousand cities; The ecosystem reached 160 thousand restaurants and 170 thousand registered deliverers working on the platform; iFood reaches more than 30.6 million orders per month.

The Technology: 

Amazon SageMaker; Amazon EKS; Amazon Kinesis and Amazon DynamoDB.

Disruptive Factor: 

SLA increased and the downtime for motorcyclists were reduced

Today, all models tested by the iFood Academy of Artificial Intelligence are already on the air with a benchmark. As a result, the company has been achieving significant gains in productivity and improvements in its service levels. According to Caetano, since the algorithms came into use, the delivery SLA has risen from 80% to 95%.

In the logistics area, the distance has been reduced by 12% thanks to the routes optimization. The same happened with the downtime of delivery professionals, which was reduced by 50%.

“When we talk about recommendations, there was also a significant improvement in conversion”, celebrates Caetano, remembering that today it is possible for iFood to optimize lists so that they point out restaurants near the customers' home, also optimizing deliveries.

Shining Moment: 

In addition to gains in SLA and time optimization for motorcyclists, iFood had gains such as:

  • The company reached more than a thousand cities
  • The ecosystem reached 160 thousand restaurants and 170 thousand registered deliverers working on the platform
  • iFood reaches more than 30.6 million orders per month.

About iFood

Founded in 2011, iFood is the leading foodtech in Latin America maintaining operations in Brazil, Mexico and Colombia. The company works with partners with initiatives that bring together business intelligence and management solutions for the approximately 160,000 restaurants registered in more than 1,000 cities across Brazil and receives 30.6 million monthly orders.